Operations, Logistics and Supply Chain Management

This book provides an overview of important trends and developments in logistics and supply chain research, making them available to practitioners, while also serving as a point of reference for academicians. Operations and logistics are cornerstones of modern supply chains that in turn are essential for global business and economics. The composition, character and importance of supply chains and networks are rapidly changing, due to technological innovations such as Information and Communication Technologies, Sensors and Robotics, Internet of Things, and Additive Manufacturing, to name a few (often referred to as Industry 4.0). Societal developments such as environmental consciousness, urbanization or the optimal use of scarce resources are also impacting how supply chain networks are configured and operated. As a result, future supply chains will not just be assessed in terms of cost-effectiveness and speed, but also the need to satisfy agility, resilience and sustainability requirements. To face these challenges, an understanding of the basic as well as more advanced concepts and recent innovations is essential in building competitive and sustainable supply chains and, as part of that, logistics and operations. These span multiple disciplines and geographies, making them interdisciplinary and international. Therefore, this book contains contributions and views from a variety of experts from multiple countries, and combines management, engineering as well as basic information technology and social concepts. In particular, it aims to: provide a comprehensive guide for all relevant and major logistics, operations, and supply chain management topics in teaching and business practiceaddress three levels of expertise, i.e., concepts and principles at a basic (undergraduate, BS) level, more advanced topics at a graduate level (MS), and finally recent (state-of-the-art) developments at a research level. In particular the latter serve to present a window on current and future (potential) logistics innovations in the different thematic fields for both researchers and top business practitionersintegrate a textbook approach with matching case studies for effective teaching and learningdiscuss multiple international perspectives in order to represent adequately the true global nature of operations, logistics and supply chains.

129 downloads 7K Views 22MB Size

Recommend Stories

Empty story

Idea Transcript


Lecture Notes in Logistics Series Editors: Uwe Clausen · Michael ten Hompel · Robert de Souza

Henk Zijm · Matthias Klumpp  Alberto Regattieri · Sunderesh Heragu Editors

Operations, Logistics and Supply Chain Management

Lecture Notes in Logistics Series editors Uwe Clausen, Dortmund, Germany Michael ten Hompel, Dortmund, Germany Robert de Souza, Singapore, Singapore

More information about this series at http://www.springer.com/series/11220

Henk Zijm Matthias Klumpp Alberto Regattieri Sunderesh Heragu •



Editors

Operations, Logistics and Supply Chain Management

123

Editors Henk Zijm Department of Industrial Engineering and Business Information Systems University of Twente Enschede, Overijssel, The Netherlands Matthias Klumpp FOM University of Applied Sciences Essen, Nordrhein-Westfalen, Germany

Alberto Regattieri Department of Industrial Engineering University of Bologna Bologna, Italy Sunderesh Heragu Industrial Engineering and Management Oklahoma State University Stillwater, OK, USA

and University of Duisburg-Essen Essen, Nordrhein-Westfalen, Germany

ISSN 2194-8917 ISSN 2194-8925 (electronic) Lecture Notes in Logistics ISBN 978-3-319-92446-5 ISBN 978-3-319-92447-2 (eBook) https://doi.org/10.1007/978-3-319-92447-2 Library of Congress Control Number: 2018943727 © Springer International Publishing AG, part of Springer Nature 2019 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Contents

Part I 1

2

3

Introductory Chapters

Objectives, Educational Developments and Structure of the Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Matthias Klumpp, Henk Zijm, Sunderesh Heragu and Alberto Regattieri Perspectives on Operations Management Developments and Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Henk Zijm, Sunderesh Heragu, Matthias Klumpp and Alberto Regattieri Operations, Logistics and Supply Chain Management: Definitions and Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Henk Zijm, Matthias Klumpp, Sunderesh Heragu and Alberto Regattieri

Part II

3

15

27

Key Domains of Supply Chains

4

Purchasing and Supply Management . . . . . . . . . . . . . . . . . . . . . . . Holger Schiele

45

5

Manufacturing Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Henk Zijm

75

6

Marketing Concepts and Instruments in Supply Chain Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thomas Neukirchen, Oliver Gansser and Matthias Klumpp

7

97

International Trade, Global Supply Chains and Compliance . . . . . 131 Albert W. Veenstra

v

vi

Contents

Part III

Overarching Topics

8

Information Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 J. Rod Franklin

9

Actionable Sustainability in Supply Chains . . . . . . . . . . . . . . . . . . . 191 Árni Halldórsson

10 Human Resource and Knowledge Management . . . . . . . . . . . . . . . 205 Matthias Klumpp, Sascha Bioly and Thomas Neukirchen Part IV

Functions in Production and Logistics

11 Inbound Logistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Stefan Minner 12 Manufacturing Planning and Control Systems . . . . . . . . . . . . . . . . 251 Henk Zijm and Alberto Regattieri 13 Packaging Logistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 Alberto Regattieri, Giulia Santarelli and Francesco Piana 14 Outbound Logistics and Distribution Management . . . . . . . . . . . . . 305 Matthias Klumpp and Sunderesh Heragu 15 Warehousing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331 Sunderesh Heragu 16 Closed Loop Supply Chain Management . . . . . . . . . . . . . . . . . . . . 353 Erwin A. van der Laan Part V

Models for Operations, Logistics and Supply Chain Management

17 Location Analysis and Network Design . . . . . . . . . . . . . . . . . . . . . . 379 Mark S. Daskin and Kayse Lee Maass 18 Process Engineering and Optimization . . . . . . . . . . . . . . . . . . . . . . 399 Marcello Braglia, Marco Frosolini, Roberto Gabbrielli and Leonardo Marrazzini 19 Advanced Production Planning and Scheduling Systems . . . . . . . . 417 Henk Zijm and Marco Schutten 20 Stochastic Inventory Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441 Henk Zijm 21 Transportation Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469 Wouter van Heeswijk, Martijn Mes and Marco Schutten

Contents

vii

22 Maintenance Service Logistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493 Joachim Arts, Rob Basten and Geert-Jan van Houtum Part VI

New Developments and Special Topics

23 Additive Manufacturing and Its Impact on the Supply Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521 Henk Zijm, Nils Knofius and Matthieu van der Heijden 24 Future Technologies in Intralogistics and Material Handling . . . . . 545 Kai Furmans, Zäzilia Seibold and Andreas Trenkle 25 Supply Chain Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575 Gerwin Zomer 26 Trends in E-commerce, Logistics and Supply Chain Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593 Gregor Sandhaus 27 Multi-agent Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 611 Martijn Mes and Berry Gerrits 28 Artificial Intelligence Applications . . . . . . . . . . . . . . . . . . . . . . . . . . 637 Matthias Klumpp 29 Advanced Green Logistics Strategies and Technologies . . . . . . . . . 663 Tim Gruchmann 30 Automatic Identification Technology . . . . . . . . . . . . . . . . . . . . . . . . 687 Michael ten Hompel, Mojtaba Masoudinejad, Omar Bousbiba and Moritz Roidl 31 The Physical Internet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 719 Eric Ballot

Editors and Contributors

About the Editors Henk Zijm is a full professor in Production and Supply Chain Management at the Department of Industrial Engineering and Business Information Systems at the University of Twente since 1990. Previously, he has been project manager at Philips Electronics in Eindhoven, and professor in Operations Management at the Eindhoven University of Technology. At the University of Twente, he served among others as Director of the Centre for Telematics and Information Technology, Dean of the Faculty of Electrical Engineering, Mathematics and Computer Science, and Rector Magnificus (Vice Chancellor). Professor Zijm is also a past president of ISIR (the International Society for Inventory Research, Budapest). In 2010, he was appointed as scientific director of the Dutch Institute for Advanced Logistics (DINALOG), a national institute responsible for executing the Dutch government imposed innovation program on logistics and supply chain management. Between 2014 and 2016 he also served as vice-chair of the European Technology Platform for Logistics, which helps to design roadmaps that drive the Horizon 2020 research and innovation program in Transport and Logistics.Professor Zijm has published more than 120 articles in international refereed scientific journals and is the (co-) author of three books. He has been a consultant to a wide variety of industrial organisations in the Netherlands and in Europe. Matthias Klumpp is a full professor in logistics at FOM University of Applied Sciences Essen and research group leader in production and logistics at the University of Duisburg-Essen and the Fraunhofer Institute for Material Flow and Logistics in Dortmund (Germany). His research is addressing primarily topics regarding artificial intelligence and digital business concepts, sustainability as well as qualification and training in the supply chain and logistics field. He has held several visiting and policy positions e.g. at the Dalle Molle Institute for Artificial Intelligence (University of Lugano, Switzerland), University of Twente (Netherlands), European University Institute (Florence School of Regulation) and

ix

x

Editors and Contributors

for the ESCO Expert Group on Qualification in Logistics for the European Commission (Brussels, Belgium). Alberto Regattieri is a full professor in Logistics and Head of the Management Engineering Master Course at the University of Bologna (Italy). His current research interests include the optimal design of manufacturing systems, innovative approaches to design and manage Supply Chains, Industrial Logistics, control and maintenance of industrial plants. He is/was responsible of several research projects in co-operation with - and funded by - the European Commission, private and public companies, universities and international research centers regarding supply chain and logistics fields. He published more than 170 scientific papers. Sunderesh Heragu is Regents professor and head of the School of Industrial Engineering and Management at Oklahoma State University where he holds the Donald and Cathey Humphreys Chair. Previously, he was the Duthie Chair in Engineering Logistics and Director of the Logistics and Distribution Institute (LoDI) at the University of Louisville. He has also served as Professor at Rensselaer Polytechnic Institute, Assistant Professor in State University of New York, Plattsburgh, and held visiting appointments at: State University of New York, Buffalo; Technical University of Eindhoven, the Netherlands; University of Twente, the Netherlands; and IBM’s Thomas J. Watson Research Center in Yorktown Heights, NY. He is the author of Facilities Design (now in its 4th edition) and has authored or co-authored over two hundred and fifty articles. He has served as Principal investigator or co-investigator on research projects totaling over $20 million funded by federal agencies such as the Department of Homeland Security, National Science Foundation, Defense Logistics Agency and private companies such as General Electric. Dr. Heragu is a Fellow of the Institute of Industrial and Systems Engineers (IISE) and has received IISE’s highest research award, the David F. Baker Distinguished Research award, as well as IISE’s Award for Technical Innovation in Industrial Engineering.

Contributors Joachim Arts is an associate professor at the University of Luxembourg. His research interests are in the application of operations research to problems in maintenance, inventory, and logistics. He often collaborates with industrial partners such as the Netherlands Railways, ASML, and Philips. He was a visiting scholar at the MIT Sloan School of Management and the MIT Operations Research Center. He is a recipient of the EURO doctoral dissertation award and a Prins Bernhard Cultuurfonds Fellow. Eric Ballot is a full professor in Production and Supply Chain Management at MINES ParisTech—PSL. He is also director of the Centre de Gestion Scientifique since 2017, visiting professor at the University of Hong Kong since 2015 and holds many auxiliary positions, including Scientific President of the French urban logistics research program, and vice-chair working group of the European

Editors and Contributors

xi

Technology Platform for Logistics (Brussels). His research interest is about supply chain collaboration and efficiency, and he is one of the three cofounders of the Physical Internet concept. Rob Basten is an associate professor at the Eindhoven University of Technology. His research focuses on after sales services for capital goods, e.g., maintenance optimization and spare parts inventory control. He is especially interested in using new technologies such as 3D printing for such services. He is further active in behavioral operations management, trying to understand how people and decision support systems can co-create value. Research is often performed in cooperation with industry. Rob has been a visiting scholar at the University of Texas at Austin and Rensselaer Polytechnic Institute. Sascha Bioly works at the Forschungszentrum Jülich and is in charge of project sponsoring for the project executing organization ETN (translated: energy technology sustainability) on behalf of the state of North Rhine-Westphalia. He is responsible for funding numerous research and development projects in the field of mobility and logistics in the OP ERDF and coordinates the subsidies made available by the EU and various ministries of North Rhine-Westphalia. Omar Bousbiba holds a PhD degree in computer science from the University Duisburg-Essen and works at Kostal Dortmund as a software developer on complex electronic control devices. Marcello Braglia graduated (with distinction) in 1988 in Electronic Engineering at Politecnico di Milano. Since 1995, he has been a Researcher in Mechanical Technology and Production Systems at the Università degli Studi di Brescia. Since 1998, he has been employed as a researcher and since 2002, as a Full Professor, in Industrial Plants at the Università di Pisa. His research activities mainly concern maintenance, reliability, production planning, lean production, logistics and statistical quality control. He is the author of about 180 technical papers published in national and international journals and conference proceedings. Mark S. Daskin is the department chair in the Industrial and Operations Engineering Department at the University of Michigan.He is the author of two books and over 80 refereed papers. He is a fellow of INFORMS and IISE and has served as the editor-in-chief of Transportation Science and IIE Transactions. He was elected to the US National Academy of Engineering in 2017. J. Rod Franklin is the Managing and Academic Director of Executive Education and Adjunct Professor of Logistics at the Kuehne Logistics University. In his roles Prof. Franklin oversees the development of academic programs for the University’s Executive Education organization, teaches classes in innovation, operations, decision and data analysis, game theory, mathematical modeling and critical thinking to graduate students and logistics professionals. His research focuses on supply chain efficiency, sustainable operations, data analytics, the Physical Internet and application of digital technologies to supply chain management. Prof. Franklin’s

xii

Editors and Contributors

research has received funding from the European Union, industry, other universities and international logistics organizations. Marco Frosolini graduated in in Mechanical Engineering at the Università di Pisa and obtained a PhD in Mechanical Engineering in 2005. He is currently a researcher in Industrial Plants within the same University. His research activities mainly concern equipment maintenance and reliability, production planning, logistics and project management. He is also interested in industrial information systems. He is author of more than 30 technical papers, published in national and international journals and conference proceedings. He is a member of AIDI (National Association of Academicians on Industrial Plants). Kai Furmans is the head of the Institute for Material Handling and Logistic Systems (IFL) and full professor at the Karlsruhe Institute of Technology (KIT). He also served as vice-dean of the department of mechanical engineering from 2006– 2015. Among other functions, he currently is chairman of the scientific advisory board of the BVL (the German Logistics Association). Roberto Gabbrielli graduated with honours in Mechanical Engineering, specializing in Energy, at the University of Pisa (Italy). He has a PhD in Energy Power Systems and is Associate Professor in Industrial Systems Engineering at the Department of Civil and Industrial Engineering of the University of Pisa (Italy). His current research concerns production planning and control, development of decision support systems for industrial investments, occupational safety, energy saving, energy storage and the reduction of the environmental impact in industrial systems. He authored more than 25 papers published in international scientific journals. Oliver Gansser is deputy director of the Institute for Empirical and Statistical Studies (IFES) at the FOM University of Applied Sciences, Study Program Economics and Management, in Munich. He is also a full professor in marketing at the same university and a member of the board of Access Marketing Management (AMM) e.V. His research focuses on behavioral typologies, preference research, communication research, and consumer behavior as well as customer relationship management. Berry Gerrits is a PhD candidate within the department of Industrial Engineering and Business Information Systems at the University of Twente, The Netherlands. He holds an MSc in Industrial Engineering (2016). His research interests are transportation, multi-agent systems, agent-based simulation and supply chain management. He is also the founder of Distribute and co-founder of DistributIT. Tim Gruchmann is a PhD candidate at the University of Kassel and holds an academic position at FOM University of Applied Sciences. His research focuses on sustainability and dynamics in supply chains, in particular in the food and automotive sector. Earlier, he has gained relevant professional experience within the automotive industry while being responsible for numerous intra- and outbound

Editors and Contributors

xiii

logistics projects in Germany and abroad (Hungary, Turkey) for various just-in-time supplier modules. Árni Halldórsson is professor of Supply Chain Management at the Division of Service Management and Logistics, Chalmers University of Technology, Sweden. Árni also holds a position as Professor of Supply Chain Management and Social Responsibility, Hanken School of Economics, Finland. Prior to this, he was at Copenhagen Business School, Denmark, where he completed his PhD, and at the University of Southampton, UK. His current research and teaching evolves around sustainable supply, energy efficiency in the supply chain, sustainable development of logistics services, service management, and theorizing and methodology in these fields. Sunderesh Heragu is Regents professor and head of the School of Industrial Engineering and Management at Oklahoma State University where he holds the Donald and Cathey Humphreys Chair. Previously, he was the Duthie Chair in Engineering Logistics and Director of the Logistics and Distribution Institute (LoDI) at the University of Louisville. He has also served as Professor at Rensselaer Polytechnic Institute, Assistant Professor in State University of New York, Plattsburgh, and held visiting appointments at: State University of New York, Buffalo; Technical University of Eindhoven, the Netherlands; University of Twente, the Netherlands; and IBM’s Thomas J. Watson Research Center in Yorktown Heights, NY. He is the author of Facilities Design (now in its 4th edition) and has authored or co-authored over two hundred and fifty articles. He has served as Principal investigator or co-investigator on research projects totaling over $20 million funded by federal agencies such as the Department of Homeland Security, National Science Foundation, Defense Logistics Agency and private companies such as General Electric. Dr. Heragu is a Fellow of the Institute of Industrial and Systems Engineers (IISE) and has received IISE’s highest research award, the David F. Baker Distinguished Research award, as well as IISE’s Award for Technical Innovation in Industrial Engineering. Matthias Klumpp is a full professor in logistics at FOM University of Applied Sciences Essen and research group leader in production and logistics at the University of Duisburg-Essen and the Fraunhofer Institute for Material Flow and Logistics in Dortmund (Germany). His research is addressing primarily topics regarding artificial intelligence and digital business concepts, sustainability as well as qualification and training in the supply chain and logistics field. He has held several visiting and policy positions e.g. at the Dalle Molle Institute for Artificial Intelligence (University of Lugano, Switzerland), University of Twente (Netherlands), European University Institute (Florence School of Regulation) and for the ESCO Expert Group on Qualification in Logistics for the European Commission (Brussels, Belgium). Nils Knofius is a doctoral student in the Department of Industrial Engineering and Business Information Systems at the University of Twente. He also received his Master’s degree in industrial engineering and management from the University of

xiv

Editors and Contributors

Twente, while holding a Bachelor’s degree from the Nordakademie. His research interest lies in after-sales service supply chains and new technologies such as additive manufacturing and artificial intelligence. His PhD research focuses on additive manufacturing (3D printing), in which he developed and evaluated various business concepts for its implementation in industry. Kayse Lee Maass is an assistant professor in the Mechanical and Industrial Engineering Department at Northeastern University. She holds a PhD from the University of Michigan, where she was a recipient of the NSF Graduate Research Fellowship Program Award. Her primary research interests lie in the application of operations research methods to location analysis, social justice, and healthcare. Leonardo Marrazzini graduated in 2016 in Mechanical Engineering at the University of Pisa. In the same year, he began his PhD studies in Industrial Plants within the same university. His research activities mainly concern the adaptation of Lean Manufacturing principles to engineering-to-order (ETO) production environments. His research focuses on models and techniques to support various company operations. Mojtaba Masoudinejad received a BSc degree in control engineering from K.N. Toosi University of Technology, Tehran, Iran, in 2007 and an MSc degree in automation and robotics from the TU Dortmund, Germany, in 2011, where he is currently working on a PhD study in the Chair of Materials Handling and Warehousing. Since 2012, he has also been a Research Assistant with the TU Dortmund. His research mainly focuses on the development of sustainable embedded and Internet of Things (IoT) devices in the field of materials handling in cooperation with the Fraunhofer Institute for Material Flow and Logistics. Martijn Mes is Associate Professor within the department of Industrial Engineering and Business Information Systems at the University of Twente, The Netherlands. He holds an MSc in Applied Mathematics (2002) and a PhD in Industrial Engineering and Management at the University of Twente (2008). His research interests are transportation, multi-agent systems, stochastic optimization, discrete event simulation, and simulation optimization. Stefan Minner is a full professor in Logistics and Supply Chain Management at the School of Management, Technische Universität München. He is the Editor-in-Chief of the International Journal of Production Economics. His research was published in many journals, including Manufacturing & Service Operations Management, Operations Research, Production and Operations Management, and the European Journal of Operational Research. He is a fellow of the International Society for Inventory Research (ISIR) and is currently vice-chairman of the scientific advisory board of the German Logistics Association (BVL), a member of the Research Committee of the European Logistics Association (ELA), and the speaker of the research training group Advanced Optimization in a Networked Economy (AdONE) at TUM.

Editors and Contributors

xv

Thomas Neukirchen is a researcher in Logistics and Service Management at FOM University of Applied Sciences, currently associated with the project MARTINA, which is aimed at research and software development for mobile logistics education and training. He studied Economics (University of Cologne, WU Vienna) with a focus on quantitative analysis and game theory. Professional experience includes event management for international conferences and quality management for an automotive supplier. Francesco Piana is an industrial engineer and PhD graduate at the Department of Industrial Engineering of Alma Mater Studiorum – University of Bologna. His favorite theme of research during the doctorate was humanitarian logistics; he studied the introduction of innovative mechanical plants in developing regions and in humanitarian camps, including waste management plants, water desalination plants and renewable energy plants. Since five years he is working at the Department of Industrial Engineering in both research and education, and is involved in European projects and collaboration with private companies. In 2018 he also started a private company based consulting activity. Alberto Regattieri is a full professor in Logistics and Head of the Management Engineering Master Course at the University of Bologna (Italy). His current research interests include the optimal design of manufacturing systems, innovative approaches to design and manage Supply Chains, Industrial Logistics, control and maintenance of industrial plants. He is/was responsible of several research projects in co-operation with - and funded by - the European Commission, private and public companies, universities and international research centers regarding supply chain and logistics fields. He published more than 170 scientific papers. Moritz Roidl works as the Chief Engineer of the Chair of Materials Handling and Warehousing at the TU Dortmund. He holds a degree in computer science (Dipl.Inform.) from the same university. His research topics include cyber-physical systems in logistics, Industry 4.0, and decentralized control of material flow systems. Gregor Sandhaus is a full professor for information systems at the FHDW University of Applied Sciences in Mettmann, Germany. His research areas in logistics are information systems for SCM and data analytics with neural networks. Prior to his professorship Gregor Sandhaus has held different management positions, e.g. as global alliance manager at Sybase or as managing director at Entercom Consulting GmbH. Giulia Santarelli is a logistic manager in a multinational corporation. She worked in the Department of Industrial Engineering at the University of Padua for a few years, where she obtained her PhD in 2013, with a study on innovative solutions for packaging. At the University she became an expert in innovative packaging solutions for humanitarian organizations. Her networks have enabled her to build up a strong partnership between the University of Bologna and the UNHRD-WFP, one of the most important humanitarian agencies in the world.

xvi

Editors and Contributors

Holger Schiele is a full professor of Innovation of Purchasing, Production and Logistics at the University of Twente in Enschede (the Netherlands). He received both his PhD and habilitation from the Leibniz University of Hanover (Germany). Previous to fully joining academia he has worked for ten years in industry and as a consultant. Twente is one of the few public universities in Europe to host a master track in purchasing and supply management, which is co-ordinated by Holger. In terms of research, Holger is one of the most prolific writers in his field. His research interests focus on innovation from and with suppliers, supplier satisfaction, supply risk management and Industry 4.0 applications in purchasing. Marco Schutten is an associate professor in the Department of Industrial Engineering and Business Information Systems at the University of Twente, the Netherlands. He holds an MSc in Applied Mathematics (1992) and a PhD in Industrial Engineering from the University of Twente (1996). His research interests include planning and scheduling in transportation, project, and production environments. Zäzilia Seibold received a PhD in Mechanical engineering in 2016 from the Karlsruhe Institute of Technology (KIT). For her thesis, she was awarded the prize for the best thesis in logistics by BVL, also in 2016. She now works at Bosch in Logistics Innovations. Michael ten Hompel is a full professor and holds the Chair of Materials Handling and Warehousing at the TU Dortmund. At the same time, he is Managing Director of the Fraunhofer Institute for Material Flow and Logistics and Director of the Fraunhofer Institute for Software and Systems Engineering, both in Dortmund. He is also board member of BITKOM, member of Acatech (National Academy of Science and Engineering), BVL and the Logistics Hall of Fame, and founding member of WGTL (Scientific Association of Technical Logistics). He published more than 400 papers and more than a dozen books. Andreas Trenkle is the head of the group Robotics and Interactive Systems in Logistics at the IFL of the Karlsruhe Institute of Technology (KIT). For his work in AGV-Systems and Interactive Systems he has won several awards. Matthieu van der Heijden is associate professor at the University of Twente, Enschede, the Netherlands. His expertise is operations research models for decision support in maintenance, inventories and transportation. Especially he focuses on after-sales service supply chains for advanced capital goods. He participated in and managed 12 research projects in close collaboration with academic and industrial partners in the area of transportation and after-sales service supply chains. He has published over 50 papers in ISI journals. Erwin A. van der Laan is an Associate Professor at RSM, Erasmus University, where he teaches operations management and logistics and heads the Center for Closed Loop Supply Chains. His research interests include closed loop supply chain management, operations of sustainable supply chains and humanitarian logistics. His work has been published in Management Science, Operations Research,

Editors and Contributors

xvii

Production and Operations Management, and European Journal of Operational Research, among others. Wouter van Heeswijk is a researcher at the Center for Mathematics and Computer Science (CWI) in Amsterdam. He obtained a master’s in Financial Engineering & Management and a PhD in Operations Research at the University of Twente. For his PhD thesis, he studied both planning techniques and transportation management in the context of urban freight logistics. Geert-Jan van Houtum is a professor of Maintenance and Reliability at the Department of Industrial Engineering & Innovation Sciences at the Eindhoven University of Technology. He also serves as vice-dean in this department. His research is focused on spare parts management, maintenance optimization, exploiting real-time degradation data for maintenance, and the effect of design decisions on the total cost of ownership of capital goods. Together with Bram Kranenburg, he wrote a book entitled Spare Parts Inventory Control under System Availability Constraints. Albert W. Veenstra is a full professor of International Trade Facilitation and Logistics at the Technical University Eindhoven. He is also the scientific director of TKI Dinalog. Albert Veenstra is involved in the agenda setting for research and innovation for the logistics industry in the Netherlands, as part of the Dutch government initiative to promote the Topsector Logistics. He has published on topics in international trade, global logistics, ocean shipping, and customs and supply chain compliance. Henk Zijm is a full professor in Production and Supply Chain Management at the Department of Industrial Engineering and Business Information Systems at the University of Twente since 1990. Previously, he has been project manager at Philips Electronics in Eindhoven, and professor in Operations Management at the Eindhoven University of Technology. At the University of Twente, he served among others as Director of the Centre for Telematics and Information Technology, Dean of the Faculty of Electrical Engineering, Mathematics and Computer Science, and Rector Magnificus (Vice Chancellor). Professor Zijm is also a past president of ISIR (the International Society for Inventory Research, Budapest). In 2010, he was appointed as scientific director of the Dutch Institute for Advanced Logistics (DINALOG), a national institute responsible for executing the Dutch government imposed innovation program on logistics and supply chain management. Between 2014 and 2016 he also served as vice-chair of the European Technology Platform for Logistics, which helps to design roadmaps that drive the Horizon 2020 research and innovation program in Transport and Logistics. Professor Zijm has published more than 120 articles in international refereed scientific journals and is the (co-) author of three books. He has been a consultant to a wide variety of industrial organisations in the Netherlands and in Europe.

xviii

Editors and Contributors

Gerwin Zomer is a senior researcher in Logistics and Supply Chain Security at The Dutch Institute for Applied Scientific Research (TNO). In TNO, he is responsible for the portfolio of innovation projects in Data Driven Logistics. He coordinated some international research projects in the field of supply chain security, such as CASSANDRA and CORE. Gerwin held several related auxiliary positions, including Advisory Board member of the Dutch Trade Compliance Community, Program Developer of the Dutch Institute for Advanced Logistics (DINALOG), and working group member of the European Technology Platform for Logistics (Brussels). He published several papers and is a sought-after speaker on supply chain innovation events and conferences.

Part I

Introductory Chapters

Chapter 1

Objectives, Educational Developments and Structure of the Book Matthias Klumpp, Henk Zijm, Sunderesh Heragu and Alberto Regattieri

Abstract Operations, logistics, and supply chains are essential enablers in a modern economy. At the same time, operational, logistics, and supply chain systems are changing fundamentally as a result of technological and societal developments, requiring both researchers as well as skilled professionals to rethink them and to incorporate new functionalities. This in turn poses new demands on the way vocational and academic learning, as well as on-the-job training programs for operations, logistics, and supply chain management are designed. This book is an attempt to serve students, researchers, and business practitioners by providing information and background material at various levels. In this introductory chapter, we discuss the scope and structure of the book. Reading this chapter is recommended to determine how to use the book in classroom lectures and seminars, gather background information or for studying specific topics. In each chapter, suggestions for further reading and resources for the growth of lateral and critical thinking are provided in order to spread the learning curve—sometimes even slightly beyond the operations, logistics, and supply chain management domain. Readers are encouraged to explore the additional material for their own development and to build general learning and research capacities. M. Klumpp (B) FOM University of Applied Sciences and University of Duisburg-Essen, Essen, Nordrhein-Westfalen, Germany e-mail: [email protected] H. Zijm Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, Overijssel, The Netherlands e-mail: [email protected] S. Heragu Industrial Engineering and Management, Oklahoma State University, Stillwater, OK, USA e-mail: [email protected] A. Regattieri Department of Industrial Engineering, University of Bologna, Bologna, Italy e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_1

3

4

M. Klumpp et al.

1.1 Preliminaries This is a comprehensive textbook on operations, logistics, and supply chain management. It reflects the many technological and societal advances that have shaped the field, will shape future developments, and is based on the knowledge and experience of a carefully selected group of experts. It attempts to be complete in addressing a broad range of topics. Each chapter is discussed at a basic, advanced and state-of the-art level, to support the learning of topics and information about them. The book is intended for a readership that on the one hand includes students within the general study subject areas of economics and management, engineering as well as geography and political studies with a focus on transportation and logistics. The addressed subject fields are operations management, logistics management, supply chain management as well as subdomains such as transportation, warehousing, inhouse logistics or the impact of new technologies, to name a few. The study levels encompass the full range of bachelor, master and Ph.D. study programs; in particular, we expect also researchers to find material that may serve as a starting point for further studies. On the other hand, the book is also designed to provide a “look-up” functionality for business professionals. Operations, logistics, and supply chain management concepts as well as state of the art information regarding the latest developments in research are included, especially in the last two sections of the book. Furthermore, many other potential readers including researchers in related engineering or management domains and administration employees are encouraged to use this book as a first glance into the world of operations, logistics, and supply chain management—a professional field that literally makes an interconnected world move and function every hour of every day in our lives. While logistics makes available products and services in our economies, it faces a Herculean task which not only has to be performed cost-effectively but also within stringent environmental and social constraints. Therefore, an interested reader may find the concepts and discussions behind the processes, people, and businesses living up to this task fascinating—or even inspiring. For many, the comfort of being able to buy a product or service at a moment’s notice is an astonishing feature of the modern world that cannot be sufficiently appreciated. In this, the authors also want to share and communicate the inspiration and motivation of dealing with professional operations, logistics, and supply chain management with students, researchers, and practitioners around the world.

1.2 Developments in Education and Teaching The academic and education landscape has changed fundamentally in the last two decades. Today, universities compete in an international, global environment—not only on the basis of research and reputation (global university rankings) but also in terms of students, business cooperation, and knowledge transfer (including spin-

1 Objectives, Educational Developments and Structure of the Book

5

offs). Additionally, the traditional learning management via curriculum development and definition of learning content has evolved to learning outcome orientation and competence testing, as reflected by the PISA and AHELO test series from the OECD.1 For Europe, the major development was the so-called “Bologna process” with the standardization of three study cycles within all European countries—and in many non-EU countries. Following suit, the European Qualifications Framework (EQF) has also changed the way vocational, academic, and continuing/on-the-job education is shaped and evaluated, i.e., the transfer of qualifications and competences is easier. In this capacity, this development is seen as a blueprint for other countries and regions. It also implies that many education and learning processes are targeted more towards outcomes and personal competence acquisition, starting with school, vocational and academic education and reaching far through life-long-learning (LLL) into all continuing education aspects. Because of these trends in Europe, the so-called “Dublin descriptors” have been developed to describe expected learning outcomes for different study levels—abstracted from the disciplinary context. Figure 1.1 provides an overview of the various descriptors, which suggest a translation towards the operations, logistics, and supply chain management domain. Another learning taxonomy is provided by Bloom (Fig. 1.2), outlining in subsequent levels the expected competences of graduates. In this case it is expected that bachelor students gain competences from the third (“apply”) towards the fourth (“analyze”) level during their studies; master students mature from the “analyze” towards the “evaluate” level during their studies, and PhD students arrive at the “create” level. As with the Dublin descriptors, the overall objective of the taxonomy is to guide the didactics and expectations of lecturers as well as the self-orientation and expectation of students.

1.3 Objectives of This Textbook Operations and logistics are cornerstones of modern supply chains, which in turn are essential in global business and economics. The composition, character, and importance of supply chains and networks are rapidly changing, due to technological innovations such as Information and Communication Technologies, Sensors and Robotics, Internet of Things, Additive Manufacturing, and Cyber Physical Systems (often referred to as Industry 4.0). Societal developments such as environmental consciousness, urbanization, and optimal use of scarce resources also affect the way supply chain networks are configured and operated. As a result, future supply chains will not just be assessed in terms of cost-effectiveness and speed, but also the need to satisfy agility, resilience, and sustainability requirements. To face these challenges, an understanding of the basic as well as more advanced concepts and recent innovations are essential in building competitive and sustain1 Cp.

http://www.oecd.org/edu/skills-beyond-school/ahelo-main-study.htm.

6

M. Klumpp et al. Cycle

Knowledge and understanding:

1 (Bachelor)

[is] supported by advanced text books [with] some aspects informed by knowledge at the forefront of their field of study

2 (Master)

provides a basis or opportunity for originality in developing or applying ideas often in a research* context

3 (Doctorate)

[includes] a systematic understanding of their field of study and mastery of the methods of research* associated with that field

Applying knowledge and understanding: 1 (Bachelor)

[through] devising and sustaining arguments

2 (Master)

[through] problem solving abilities [applied] in new or unfamiliar environments within broader (or multidisciplinary) contexts

3 (Doctorate)

[is demonstrated by the] ability to conceive, design, implement and adapt a substantial process of research* with scholarly integrity

Making judgements: 1 (Bachelor)

[involves] gathering and interpreting relevant data

2 (Master)

[demonstrates] the ability to integrate knowledge and handle complexity, and formulate judgements with incomplete data

3 (Doctorate)

[requires being] capable of critical analysis, evaluation and synthesis of new and complex ideas

Communication: 1 (Bachelor)

[of] information, ideas, problems and solutions

2 (Master)

[of] their conclusions and the underpinning knowledge and rationale (restricted scope) to specialist and non-specialist audiences (monologue)

3 (Doctorate)

with their peers, the larger scholarly community and with society in general (dialogue) about their areas of expertise (broad scope) Learning skills:

1 (Bachelor)

have developed those skills needed to study further with a high level of autonomy

2 (Master)

study in a manner that may be largely self-directed or autonomous

3 (Doctorate)

expected to be able to promote, within academic and professional contexts, technological, social or cultural advancement

Fig. 1.1 Dublin descriptors

1 Objectives, Educational Developments and Structure of the Book

7

Fig. 1.2 Bloom’s taxonomy

able supply chains and, as part of that, logistics and operations. This book aims to provide an overview of important trends and developments in logistics and supply chain research, to make them available to practitioners, while also serving as a point of reference for academicians. Operations, logistics, and supply chain management span multiple disciplines and geographies, making them interdisciplinary and international. Therefore, this book contains contributions and views from a variety of experts from multiple countries, and combines management, engineering as well as basic information technology and social concepts. In particular, it aims to: • Provide a comprehensive guide for many relevant and major logistics, operations, and supply chain management topics in teaching and business practice. • Address three levels of expertise, i.e., concepts and principles at a basic (undergraduate, B.Sc.) level, more advanced topics at a graduate level (M.Sc.), and finally recent (state-of -the-art) developments at a research level. In particular the latter serve to present a window on current and future (potential) logistics innovations in multiple thematic fields for both researchers and top business practitioners. • Integrate a textbook approach with matching case studies for effective teaching and learning. • Discuss multiple international perspectives in order to adequately represent the true global nature of logistics and supply chain management.

8

M. Klumpp et al.

1.4 Structure and “How to Use” This textbook is divided into five sections for a total of 31 chapters. All chapters are further structured in three levels (basic, advanced, state-of-the-art) in order to allow for an easy understanding and clear information structure for readers and learners using this book. These levels are aligned with the bachelor, master, and Ph.D. levels in academic studies but are not limited to that. Professionals may also use this structure for guidance regarding their interest by looking up basic information in the basic sections or—say—learning about the latest developments in state-of -the-art logistics research. At the end of this chapter, there is also a comprehensive “graphical structure” included to highlight and visualize the topical structure and sequence of the book. Part I—Introductory Chapters The first three chapters of this book serve to “set the stage”. So far, we have outlined the mission and output orientation of the book, as well as some insights in the way learning and teaching has changed under the influence of learning taxonomies. In the remainder of this chapter, we present a storyline that explains the structure of the book in more detail. From Chap. 2 onwards, we focus on the content. First, we briefly review major technological and societal developments that have fundamentally changed the way logistics and supply chains are designed and managed. Disrupting business models are discussed after which we focus on the effects of the digital revolution and its impact on manufacturing and logistics. To understand why these changes are indeed revolutionary, we go back to the early roots of manufacturing, logistics, and trade (Chap. 3), present an extensive case study on the logistics operation of the Dutch East Indies Company in the 17th century, and provide definitions and concepts used throughout the book. Relations with other management disciplines and scientific domains are also reviewed. Part II—Key Domains of Supply Chains Chapters 4–7 treat the basic elements of supply chains, i.e., sourcing, making and selling, followed by a study on the role of global supply chains in international trade, with a focus on compliance. The importance of sourcing and purchasing is reflected by the fact that many companies spend more than half of their turnover on supplies. In Chap. 4, purchasing and supply management is therefore extensively discussed at multiple levels, from strategic sourcing to operational procurement, with a focus the various contexts and roles that can be observed in practice. Chapter 5 focuses on the development of manufacturing, beginning with some history that goes back more than 2,000 years, but quickly moving to the first industrial revolution and the rise of mass

1 Objectives, Educational Developments and Structure of the Book

9

production. Next, a process typology and a framework of the various manufacturing functions is discussed in detail, while providing a glimpse on future systems constantly being transformed by both technological (digitalization, robotics) and societal developments (e.g., the circular economy). Marketing concepts and instruments are discussed in detail in Chap. 6. Although the close interaction between marketing and logistics is obvious, there are not that many textbooks on logistics and supply chain that include an in-depth treatment. In this chapter, various methods to define and distinguish markets, to set up distribution channels, to select partners and to monitor and manage the key concept of time-to-market are discussed. Chapter 7 focuses on international trade and the conditions that determine to a large extent the operation of global supply chains, emphasizing national and international regulations, and the need for shippers, logistics service providers, and transport companies to comply with the rules imposed by customs and tax authorities. Special attention is given to trusted trade lanes as a next step in international supply chain compliance. Part III—Overarching Topics Chapters 8–10 discuss three topics that are fundamental for the management of logistics and supply chain operations, although their impact goes far beyond these disciplines and in fact touch almost all aspects of society. That certainly holds for information and communication technology, which is discussed in Chap. 8 in terms of its impact on modern business development. The recent emphasis on data management and data analytics and the development of new technologies such as the Internet of Things only further underlines the key role of digitization. In Chap. 9, we turn to the concept of sustainability, defined in a broad sense, i.e., social, environmental and economic sustainability (people, planet, profit). In view of the constantly occurring depletion of natural resources, but also the emission of hazardous materials, the move towards more ecological friendly logistics and supply chains is now broadly recognized. However, the key challenge will be to develop business models that reconcile economic, social, and environmental goals. Chapter 10 focuses on another fundamental condition: the need for skilled and knowledgeable workers at multiple levels. The rapid advance of new technologies, in the future more and more equipped with artificial intelligence applications, will definitely lead to an upward shift in terms of required qualifications and competences. How to prepare future workers for the world in which they have to perform is a challenging task for educational institutes and Human Resource departments. Part IV—Functions in Production and Logistics The next set of chapters discusses the sequence of steps encountered in a supply chain in more detail. Chapter 11 is devoted to inbound logistics, i.e., all activities that

10

M. Klumpp et al.

secure the supply of materials, components or products for manufacturing, assembly, and retail operations. The associated information and materials flow involve different strategic and operational decisions that influence transportation, handling, and inventories. In Chap. 12, we turn to manufacturing planning and control systems and discuss a number of production philosophies and associated control systems in more detail. Starting with the well-known Economic Production Quantity, we sketch the essentials of Manufacturing Resources Planning and Hierarchical Production Planning systems after which we turn to Just-in-Time and Lean Production concepts. This chapter is concluded with a brief discussion of digital and cloud manufacturing. A topic not often found in textbooks on supply chain management is packaging logistics, although packaging may severely influence materials handling and transport costs, as well as safety conditions. Last but not least, smart packaging may substantially reduce the environmental footprint of supply chains. These aspects are discussed in detail in Chap. 13. Outbound logistics and distribution management, the core subjects of Chap. 14, again concerns a subject of high strategic importance, especially in the trade and retail sector, and certainly in the light of ecommerce or multi-channel distribution. Basic operational concepts are introduced, after which we describe differentiations in terms of multi-echelon inventory models and multi-objective concepts, e.g., service levels, cost optimization, batch and emergency deliveries, but also a more recent phenomenon involving customer integration. A key element in any distribution concept are the warehouse and the logistics operations in warehouses, which is the subject of Chap. 15. The success of e-commerce, for example, and the ability to deliver many products within 24 hours, critically depends on, again, digital communication but above all on a superb logistics operation, beginning in the warehouse. Design concepts, as well as storage, order-picking and routing policies are discussed in depth, followed by a vision on the impact of new technologies but also mass customization and environmental concerns. The latter also provide the driving force behind the emergence of closed loop supply chains (Chap. 16) which are an operationalization of the idea of a circular economy, in which discarded products, parts or materials are reused instead of being sent to the landfill. A framework of various closed loop supply chain configurations is introduced and a number of business values are discussed. More advanced topics include the forecasting of the flow of reusable products and the integration of forward and reverse networks. Part V—Models for Operations, Logistics and Supply Chain Management Although some of the preceding chapters already include basic models to study the trade-off between various alternatives, the six chapters in Part V are primarily model-oriented and discuss optimization aspects of various functions in logistics, manufacturing, and supply chain management in detail. The sequence begins in Chap. 17, followed by a new set-covering formulation of a facility location problem in the United States. Multi-objective problems and models that integrate inventory

1 Objectives, Educational Developments and Structure of the Book

11

planning into facility location modeling are briefly outlined. Chapter 18, on process engineering and optimization, again discusses a subject not often found in operational analysis textbooks on manufacturing and supply chains. Process engineering concerns the selection of manufacturing processes and parameters and therefore largely determines the input for the subsequent manufacturing scheduling problem as encountered especially in job shops. Production planning and scheduling is also the topic of Chap. 20 in which a number of advanced models are discussed. In particular, the relationship between capacity planning and lead-time management is presented in detail, including the integration of workload control concepts. Last but not least, the shifting bottleneck heuristic for job shop scheduling is discussed. The various phases in the supply chain, but also within a manufacturing and assembly plant, are generally separated by inventory buffers. Stochastic inventory models, discussed in Chap. 20, are a rich topic in the Operations Management literature; they pertain to the analysis of anticipation stock in view of independent, and therefore uncertain, demand. Both single and multi-echelon models are discussed in a uniform analytical framework. Chapter 21 discusses transportation models in depth, with a focus on multi-modal transport and the need to coordinate and consolidate the various flows, to reconcile economic and environmental goals. This chapter includes a discussion of urban freight transport in city distribution, and the ways in which city distribution centers may help to mitigate negative effects such as hazardous particle emission and emission. Chapter 22 is primarily devoted to after-sales service logistics, specifically the logistics of spare parts and other resources in view of the maintenance of capital assets. The shift from corrective via preventive to predictive maintenance (e.g., based on condition monitoring) has a large impact on the way activities are prepared logistically, and will lead to the development of new business models, especially when responsibilities are transferred from the asset operator or owner to external service providers. Part VI—New Developments and Special Topics The last set of chapters is devoted to new and recent developments and topics that deserve special attention. Chapter 23 discusses the impact of additive manufacturing, also known as 3D-printing, on supply chain operations. It is well known that the design freedom that comes with 3D-printing marks an important step towards mass customization. Special attention is given to additive manufacturing of spare parts, in view of its perceived potential to reduce the need for spare parts inventories substantially. In Chap. 24, we return to materials handling in warehousing and intralogistics. A fundamental categorization is presented, followed by an outline of recent and future material handling concepts and functional challenges posed upon them. The topic of supply chain security, discussed in Chap. 25, is linked to the discussion on compliance to regulations that is also addressed in Chap. 7, but includes a discussion on security aspects and a treatment of business perspectives. Although already touched in several preceding chapters, e-commerce has had such a far-reaching impact on the

12

M. Klumpp et al.

consumer market that it deserves a separate treatment. Hence, this is done in Chap. 26. Both the information flows associated with on-line shopping and electronic payment) and the interdependencies between e-commerce, logistics, and supply chains are discussed, including an overview of advanced technologies. Most supply chains are heterogeneous in the sense that different parts are controlled by independent entities while business contracts govern the overall supply chain. Such agreements can be modeled by so-called multi-agent systems, which are the topic of Chap. 27. Often, such systems are also proposed for controlling manufacturing and internal logistics operations, but in particular the negotiation process leading to contractual agreements between various supply chain stakeholders. The latter is naturally modeled by multi-agent systems. Artificial Intelligence applications in logistics are the topic of Chap. 28. Relatively well-known examples include smart navigation systems and applications in autonomous vehicles but in particular the use of more advanced topics such as machine learning, data analytics, and cognitive computing may severely impact future applications. Special attention is given to man-machine interactions and the adoption rate of AI applications. Chapter 29 discusses the design of green logistics strategies and coordination in view of corporate social responsibility and the need to develop more sustainable supply chains. As mentioned previously the development of sustainable business models constitutes a major challenge to industries, governments as well as the consumer. At the same time, we have witnessed the introduction of a number of new technologies in manufacturing and logistics that help to make processes more efficient and less environmentally damaging. Chapter 30 focusses in particular on Remote Frequency Identification (RFID), which is ubiquitous in many logistics operations. Chapter 31 discusses a new paradigm in logistics and supply chain management, i.e., the Physical Internet. The Physical Internet mimics the digital internet but is devoted to the transport of physical products and materials. It assumes a basic infrastructure consisting of interconnected networks, and is based on building blocks such as modular packages (small containers), information protocols, open hubs and marketplaces as well as myopic (multi-modal) routing strategies. Ultimately, it aims at replacing the current, highly fragmented, logistics infrastructure with a system that is more efficient and at the same time more resilient. However, it represents a long-term vision; more research and experimentation are required before being realized. Figure 1.3 provides an overview of the structure of the book.

1.5 Further Reading Further insights on education and learning are provided among others by the following sources mentioned in the above topics: Regarding the new qualification framework concept and the Dublin descriptors, see European Commission (2015) for example.

13

Fig. 1.3 Overview of the book

1 Objectives, Educational Developments and Structure of the Book

14

M. Klumpp et al.

In the field of educational objectives and the Bloom taxonomy further details can be found in Anderson et al. (2001) as well as Bloom and Krathwohl (1956).

References European Commission (2015) Explaining the European Qualifications Framework, European Commission Anderson LW, Krathwohl DR, Bloom BS (2001) A taxonomy for learning, teaching, and assessing: a revision of Bloom’s taxonomy of educational objectives. Allyn & Bacon Bloom BS, Krathwohl DR (1956) Taxonomy of educational objectives book 1: cognitive domain. Longman

Chapter 2

Perspectives on Operations Management Developments and Research Henk Zijm, Sunderesh Heragu, Matthias Klumpp and Alberto Regattieri

Abstract In this chapter, we discuss major technological and societal developments that determine our lives to a large extent, not in the least the way we organize our resource needs and hence logistics and supply chains. After outlining these developments globally, we give some examples of truly disrupting business models that are fundamentally changing our logistics and supply chain operations. Next, we look in more depth at the manifestation of the digital revolution in manufacturing and supply chains, as well as the many opportunities that these technologies offer in addressing a number of major societal challenges.

2.1 Developments in Society and the Business World In general, five main trends are driving our societies and the business world in the 21st century: • Technological developments are a major driving force as for example the introduction of the smartphone in 2007 has shown. Within the industry, such trends encomH. Zijm (B) Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, Overijssel, The Netherlands e-mail: [email protected] S. Heragu Industrial Engineering and Management, Oklahoma State University, Stillwater, OK, USA e-mail: [email protected] M. Klumpp FOM University of Applied Sciences and University of Duisburg-Essen, Essen, Nordrhein-Westfalen, Germany e-mail: [email protected] A. Regattieri Department of Industrial Engineering, University of Bologna, Bologna, Italy e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_2

15

16









H. Zijm et al.

pass additive manufacturing, RFID and Internet of Things based logistics, smart robotics in warehousing and manufacturing/assembly, and cloud-based planning and control systems. Societal trends as for example the increasing individualization and customization are a major force in market development, as customers increasingly demand tailored products and services. Next to product price and quality, speed of delivery is a major distinguishing business characteristic; indeed smart logistics is an indispensable competitive field of the rapidly growing e-commerce sector. Globalization is continuing further, integrating an ever larger part of the world into more complex relations (social, financial, political), including global supply chains. This increases the task of operations, logistics and supply chain management in the design, control and resilience management of supply chains. Furthermore, claims of justice and environmental awareness (“one planet”) are voiced increasingly in the wake of globalization; fascinatingly, this concern recently induced steps towards a “local for local” economy again. Volatility and fragility is increasing: as the global recession of 2008 has shown and as is obvious from the different environmental and climate risks, humanity faces increasing crisis and risk situations. In terms of agility and robustness, also business models and supply chains have to live up to these challenges. Sustainability is a pervasive concept that is introducing significant modifications in products and industrial processes. Modern supply chains must pay attention to footprint and to the end or life of mass consumer products. More and more, authorities are putting restrictions on the movement of goods due to environmental limits.

2.2 Disruptors and Their Impact on Supply Chains Beginning with the proliferation of the internet, the economy as a whole and operations management in particular, have seen a variety of dramatic changes. The rate of change as well its pace appears to be only be accelerating. In his book titled “The World is Flat: A Brief History of the Twenty-First Century”, Friedman (2005) discusses how ten flatteners have leveled the playing field from companies and competitors across the globe. Services can be performed seamlessly at distant locations in different time zones in such a way that it can be rendered to a customer located thousands of miles without any delay. Technology has revolutionized how retail giants such as WalMart streamline procurement, shipping, distribution, and sales. It has allowed companies to provide value-added services beyond their core competency. For example, in its Louisville facilities, UPS was able to perform repair of Toshiba notebooks in addition to shipping and distribution. In the past ten years, the pace of change has only accelerated. Companies that are household names and with market capitalization of a billion or more US dollars did not exist until just a few years ago. For example Facebook, with a market capitalization of $523 billion, was founded only in 2004. Uber, valued at more than $65

2 Perspectives on Operations Management Developments and Research

17

billion was founded in 2009. Netflix, founded in 1997 as a DVD sales and rental company, has now an entirely different business model. It recently has transformed into a different company that offers movies on-demand and is valued at $121 billion. Tesla, a $56 billion maker of electric vehicles was founded in 2003. Airbnb, valued at $31 billion was founded in 2008. The list goes on and on. What do the market valuations and company-founding dates tell us? With a new business model, companies can become dominant players in their industry segments in no time. Consider the following: • Uber is the world’s largest taxi company, but owns no cars. • Netflix is the world’s largest movie house, but owns no theaters. • Apple and Google are the largest software vendors, yet they do not develop their own applications (apps). • Alibaba is the world’s largest retailer, but carries no inventory. • Skype is one of the largest phone companies, but has no physical infrastructure. • Facebook is the world’s largest content provider, but develops no content on its own. Yet, the impact of such companies on the economy, society (e.g., Arab Spring), and governments (role of Russian bots in the 2016 US elections), are of a scale and scope that it may take many years to understand their true effect.

2.3 Digitization in Logistics and Supply Chain Operations The digital transformation has revolutionized both businesses and personal life in recent decades. Along with that, we witness an unprecedented growth in the production and usage of data, largely generated automatically, through smart sensor and monitoring systems and less by human activities, although the underlying algorithms are still human designed. The success of companies like Amazon, Google and Alibaba is in the first place due to their ability to process and analyze large amounts of data and to use the results of such analyses for the development of personalized services. But many, if not all, operational processes are heavily touched by digitalization. Below, we discuss a number of direct consequences of digitalization on the way we organize our logistics and supply chain processes.

2.3.1 Supply Chain Transparency, Safety, and Security Increased data transparency may help to improve safety and security in supply chains, particularly to diminish the risks of fraud. Electronic document handling is already fully engrained in commerce. The goal for each stakeholder in a complex, heterogeneous supply chain is to be able to verify all the required information, to access and modify the data to the extent authorized, and to ensure a secure and uninterrupted

18

H. Zijm et al.

logistics process such that the goods are delivered from the producer to the receiver without any delay or loss in any link in a complex chain. A promising development may be the use of blockchain technology. A blockchain basically is a growing list of records (blocks) such that each block contains an encrypted hash of a previous block which makes it inherently resistant against modification of data by design. As an open, distributed ledger it enables easy verification and possibly certification of transactions by relevant partners and therefore offers prevention against any act of theft or fraud. The study of blockchains is still in its infancy (and is unfortunately too much dominated by the bitcoin debate).

2.3.2 Supply Chain Control Towers and the Physical Internet The growing interest in supply chain control towers is a direct consequence of the enhanced transparency caused by increased data visibility. However, the translation of data into meaningful information is not self-evident. Classical data analytics techniques such as statistical analysis and pattern recognition are now enriched with tools arising in the field of artificial intelligence (see Sect. 2.4.3). Control towers may help individual companies to improve logistics and supply chain control at both a tactical and an operational level, but also serve to synchronize operations of various companies with the aim to both reduce costs and negative environmental effects, e.g., by pooling resources in freight logistics across supply chains (horizontal collaboration) or by integrating sales, manufacturing and distribution schedules along supply chains (vertical collaboration). The Physical Internet presents a vision of a common infrastructure used by logistics service providers which entirely takes over responsibility from shippers and customers for a smooth and seamlessly integrated handling of transport orders. It is similar to the file-handling system of the digital internet, and is based on standardization of modular packets, electronic documents, routing protocols and payment procedures. The Physical Internet is an attempt to overcome the current highly fragmented world of shippers, forwarders, logistics service providers, supply chain finance agents and other stakeholders that lacks pooling and synchronization.

2.3.3 E-commerce and Last-Mile Delivery The unprecedented growth and success of e-commerce transactions is largely due to a superb information infrastructure and to excellent logistics networks in which every step in the order and shipping workflow is recorded, allowing the customer to trace the order at any time and if needed to influence the shipping schedule. Next day or even same day delivery is only possible with a highly advanced logistics order handling system. Yet, the challenge will be to keep the system sustainable, in particular due to the return shipments that come along with the growth of e-commerce. Urban and last-

2 Perspectives on Operations Management Developments and Research

19

mile transport are known to be the most expensive part of a logistic distribution chain which again requires well-thought control towers focusing on horizontal and vertical collaboration. Companies can efficiently transport products (large and small) across thousands of miles on container ships, trains, and trucks, but getting it to the end customer (last-mile delivery) is costly and time consuming. There are several factors that affect the last-mile delivery, including traffic congestion, lack of infrastructure, and uncertainty in exact delivery time. Amazon offers its Prime customers two-hour delivery of groceries and many household items in thirty metropolitan cities in the US. With their recent acquisition of Whole Foods and vertical integration into the logistics business, they are able to minimize the negative experiences relative to the last-mile delivery for their customers by shipping goods to customers at a time and location specified by the customer. Driverless vehicles also offers the potential to impact the last-mile transportation. Many customers in Europe who have access to comfortable and relatively punctual high-speed train travel still prefer to use their personal vehicles for travel, in part, because of the need to get their exact locations at their destination from the train station. Imagine a future state where there is a sufficient number of driverless cars that can traverse all the areas within five miles of a train station. Drivers who today prefer to drive their own cars because of the difficulty in reaching their final destination, may now be willing to utilize driverless vehicles at the train station provided the wait times for these vehicles is minimal. In fact, this is even more attractive than driving one’s own vehicle because the customer does not have to worry about parking anymore. Today, this is already possible with Uber pools.

2.4 Industry 4.0/Smart Industry The term Industry 4.0 or Smart Industry indicates the revolutionary change that digitalization and in particular the introduction of cyber-physical systems has brought about in almost all aspects of the industrial manufacturing and services industry. Without attempting to be complete, we discuss a few of the most prominent manifestations of Industry 4.0.

2.4.1 Additive Manufacturing A promising new production technology is additive manufacturing, often referred to as 3D printing. Its name stems from the fact that it is based on adding material in thin layers, one layer at a time, to build a product. A variety of materials, including composite materials that can be engineered to have the desired mechanical and structural properties, can be used. The product’s shapes and characteristics are completely digitalized and stored in a CAD file which subsequently is translated into a digital process plan. The advantages of 3D printing are twofold. First, because the

20

H. Zijm et al.

3D-printers are small enough to be put on a desk, individual customers can print (manufacture) small, spare parts to replace those that are damaged or dysfunctional. For example, a key component of an equipment on a navy ship that is currently in the middle of the Pacific Ocean that has failed can be replaced quickly without the need for the spare part to be shipped. Second, 3D printing is based on the large degree of freedom a designer has in developing new products, as opposed to classical machining operations based on the removal of materials. As a result, 3D printing is an important step in offering highly customized, if not unique, products to customers. However, through its integration of functions, 3D printing negates the advantages of modularization, which is why some scholars are skeptical on its use. This is a topic for further study.

2.4.2 Internet of Things Sensor-technology has developed rapidly from classical condition monitoring techniques (for instance used in maintenance and service operations, and quality assurance) to Remote Frequency Identification (RFID) that allows for the identification of objects without the need of visual access to the identifier. The next step is to connect automatic signaling to subsequent actions. The term Internet of Things is used to denote devices that communicate with each other without necessary human interference and subsequently activate automatic devices to perform follow-up actions (e.g. robots in a warehouse to pick replenishment items for transport to the desired location. Another interesting field is maintenance and service management where by means of remote monitoring the need for preventive maintenance actions is detected and subsequently both logistics and repair actions are automatically planned.

2.4.3 Artificial Intelligence and Machine Learning The use of Artificial Intelligence techniques is expected to play a dominant role in many societal domains, including logistics and supply chains. While cognitive computing is often used for the processing of highly unstructured data sets (e.g., by Natural Language Processing), machine learning is a promising technique to enhance automatic decision making in more structured environments, without defining and coding all applicable rules beforehand. Machine learning may be based on a variety of techniques (e.g., neural networks, logistic regression) and can play an important role in predictive analytics. As an example, we mention the shift from preventive to predictive maintenance, based on smart condition monitoring of assets and, if needed, the automatic determination of actions to restore a desired functionality. Both cognitive computing and machine learning techniques have proven their value in the development of personalized marketing instruments but are also expected to play a key role in future operations design.

2 Perspectives on Operations Management Developments and Research

21

2.4.4 Robotics and Driverless Vehicles Automation in manufacturing and logistics has already an impressive track record (computer integrated manufacturing, robotic assembly, automatic guided vehicles, automated storage and retrieval systems) but so far concerned primarily indoor operations. The increased pace of the development of autonomous vehicles will not only impact passenger but also freight transport, both in last-mile distribution and in longhaul transport, where we now see experiments with platooning of freight trucks. Similarly, unmanned cargo aircrafts (not to be mistaken with drones) are receiving considerable attention, in particular for the transport of goods in less developed or densely populated regions. Clearly, such developments pose important new research questions in terms of the infrastructure needed to allow such autonomous devices to operate, including the use of Internet of Things technology to move safely around obstacles and to prevent collisions. Robotics, autonomous vehicles and unmanned aircraft (including drones) and ships are not just examples of high-level technical automation but in addition use artificial intelligence techniques to make short-term decisions autonomously.

2.4.5 Cloud Computing and Cloud Manufacturing Cloud Computing indicates the use of a common space and infrastructure to perform a large variety of digital operations and thereby to diminish the need of local computing capacity and exploiting shared resources and communication infrastructures. It also offers the possibility to share work around the globe, as already happens in product design for many years by means of Electronic Data Interchange (EDI). Cloud Manufacturing is actually the manufacturing equivalent of cloud computing and is a direct consequence of the possibility to define all aspects of a product or part digitally, including the ways it should be processed. The emergence of 3D-print service companies is just one manifestation, but also the use of a common infrastructure for manufacturing planning and control (cloud-based Enterprise Resource Planning systems) is usually viewed as an example of cloud manufacturing. Technologies such as 3D printing, possibly executed as a cloud manufacturing solution, enhance our possibilities to present customized solutions to an increasingly diverse client base. In some fields, a further paradigm shift is observed in that customers are primarily interested in platform-based services (music streaming, cloud computing or manufacturing, car sharing), without necessarily owning the platform itself (shared economy).

22

H. Zijm et al.

2.4.6 Virtual Reality As the virtual reality technology continues to improve, many services in business can be severely impacted. In systems design, virtual reality offers the option to explore alternative designs, e.g. in future building constructions by enabling the potential user to “walk through” a particular variant with the aim to provide useful feedback to the designer. In manufacturing and transport, virtual reality enables engineers to become familiar with new techniques or with the control of complex machinery without being physically confronted with it (as a flight simulator used for training car drivers or aircraft pilots). Serious games have proven to be an excellent tool in familiarizing management with the effects of their decisions made in complex circumstances. In maintenance we find applications in which an engineer on the spot receives instructions from a digital device that contains and displays a full model of the object under inspection. This is a field in which also augmented reality has proven to be helpful. As the term suggests, augmented reality adds computer-generated (often, graphical) information to elements observed in the real world. The information may be based on data from sensors that is not directly observable (a simple example is an electronic parking assistant in a car) but may also use background database information (an eye-glass that projects information on a person you observe in reality). Augmented images may also be entirely fictitious in which case they are often projected on real world images and observed via special equipment such as a head-mounted display. Both virtual and augmented reality have raised high expectations in training people for a variety of tasks in manufacturing and logistics, but also in helping to secure safety in logistics and supply chain operations.

2.5 Societal Developments The growing world population and the depletion of natural resources has an unmistakable impact on our living circumstances, including the supply of products and services. Urbanization is placing a heavy burden on our economic, ecological and social environment. Many people recognize the need to become more efficient in the use of resources and to diminish their ecological footprint. There are basically two ways of answering the challenge: do more with less (shared economy) and avoid any unnecessary waste. Below, we elaborate on both ways.

2.5.1 Shared Economy We briefly touched upon the need for a shared economy. Many resources today are underutilized. Consider the cars parked by customers who are taking a short trip from their local or regional airport. Until the customer returns, the vehicle is

2 Perspectives on Operations Management Developments and Research

23

not used. In addition, valuable space is occupied. Of course, the parking company receives revenue for parking and the customer is willing to pay for the convenience of parking and having access to their car immediately upon their return. However, when looking at this situation from a shared economy perspective, one might think of a business model in which the parking lot owner does not charge the car owner for parking. In return, the parking lot owner rents this car to other customers who need a vehicle during the time period that the car owner is away. This generates revenues for the parking company and some customers may be motivated to rent their car if offered free parking and potentially additional financial incentives. Another wellknown example of shared economy is how companies utilize truck capacities so that a trailer is completely filled with goods (coming from multiple customers) in the outbound and inbound trips. There are some industries where the return trip is almost empty. For example, the Scania truck plant in the Netherlands receives pallet loads of subassemblies via trucks each morning from Sweden, but these trucks take back only the empty pallets on their return trip. Warehouses can be filled to capacity if goods from multiple customers are pooled. Of course, there are some drawbacks in doing so because if a primary customer’s storage space is allocated to another customer and there is a sudden change in demand for the primary customer’s goods, there may not be sufficient space to accommodate the inventory of all the customers and the warehouse manager will have to secure additional space at a much higher cost. Third-party logistics providers use sophisticated mathematical models to manage their resources effectively. With the growth of the shared economy, there will be continued interest in this area.

2.5.2 Sustainability and the Environment: The Circular Economy Sustainable and resilient/robust global supply chains are a key condition to maintain and distribute welfare and prosperity to both developed and developing countries and hence we have to find ways to cope with these challenges. Some key figures may serve to illustrate the need to find solutions to these challenges. While the European Committee has set clear targets to reduce Greenhouse Gas Emissions (GGE) in 2015 to 60% as compared to 1990, we observe that the percentage of transport related GGE has increased from 25% in 1990 to 36% currently. Intensified global logistics has also increased security vulnerability, while both volume and speed increase have introduced additional safety risks. There are both technological and organizational ways to turn the tide. The development of cleaner and more efficient engines, the electrification of transport and the use of hydrogen power instead of fossil fuels are just one way, but the efficient use of resource capacity by pooling resources is an important additional step. Sharing of resources however often requires the collaboration of competitors who therefore have to share information. The hesitation to share competition-sensitive and hence

24

H. Zijm et al.

confidential information presents a significant obstacle for attempts to coordinate and synchronize supply chains. Another way to diminish the footprint is the re-use of products, components or materials wherever possible. New technologies for materials separation may help but also the design of so-called closed loop supply chains is a way to move from a linear to a circular economy in which materials and products are re-used instead of wasted. Industrial Symbiosis is also an attempt to combine economic and ecological goals, in which industries may use residue or waste products from sometimes entirely different industries as feedstock in their production processes. Such a symbiotic relationship may indeed reduce the charges paid for landfill, while at the same time diminish purchasing costs (of fresh materials), thereby providing another example in which economy and environment both win.

2.6 Summary and Conclusions Future research in the fields of Operations, Logistics and Supply Chain Management will undoubtedly focus on the interfaces with technologies that arise in the digital revolution, or better, to adopt the many possibilities that these new technologies offer in both the execution and the control of processes. Such innovations are badly needed in view of another development that currently threatens our lives, i.e. the consequences of climate change and the need to diminish the social and ecological footprint of logistics and supply chains, while still retaining a profitable business. The ultimate challenge will be to integrate these new technologies into business models that no longer aim at profit maximization or cost minimization of individual businesses but instead address societal, ecological and economic concerns, i.e. that are oriented towards sustainability in its full breadth (people, planet, profit). In this chapter, we have highlighted developments that we believe will have a profound impact on the way future supply chains need to be designed and operated. Technological advancements and in particular the digital revolution will unleash an unprecedented potential to improve our standards of living in a sustainable way. The ultimate challenge however will be to make it happen and that requires first and foremost the design of business models that respect and balance the interests of many partners and stakeholders involved. When acting selfish, such stakeholders may often face conflicting objectives that prevent collaboration. For instance, horizontal collaboration of logistics service providers in a supply chain requires the sharing of commercially sensitive information, which may raise resistance of companies. The transition from product sales to product-based services again requires a fundamental change in the way contracts between producers and customers are designed, aiming at a lifelong relationship rather than an isolated transaction. A key element in modeling these changes and solving potential conflicts will be to include societal costs into the economic equation. Examples include road pricing for both freight and passenger transport or alternatively incentivizing pooling, a ban on heavy trucks in inner cities, stimulating closed loop supply chains either through penalties or incentives to avoid

2 Perspectives on Operations Management Developments and Research

25

unwanted landfill, etc. Such solutions may require unorthodox policy measures or regulations, stimulating the development of new contracts between partners. Game theoretical models and methods often help to analyze and resolve potential conflicts, leading to equilibrium solutions that are acceptable for all stakeholders involve. For short-term automatic decision making, modeling the negotiation process as a multiagent systems may prove valuable. However, the ultimate value of any solution will critically depend on the constraints involved which may include legal, psychological and ethical elements, next to technological and systems engineering perspectives.

2.7 Further Reading A very readable introduction on how a number of major developments have changed the level playing field for industries is Friedman (2005). A nice overview of recent innovations is Logistics and Supply Chain Management is provided by Zijm et al. (2016). Almost all topics discussed in the preceding sections are represented in subsequent chapters in this volume.

References Friedman TL (2005) The world is flat: a brief history of the twenty-first century. Macmillan, New York, NY Zijm H, Klumpp M, Clausen U, Ten Hompel M (eds) (2016) Logistics and supply chain innovation: bridging the gap between theory and practice. In: Lecture notes in logistics. Springer International Publishing, Berlin

Chapter 3

Operations, Logistics and Supply Chain Management: Definitions and Objectives Henk Zijm, Matthias Klumpp, Sunderesh Heragu and Alberto Regattieri

Abstract Operations, logistics and supply chains are catalysts in any modern economy and therefore essential contributors to economic prosperity and societal welfare. This chapter briefly sketches the origins of the field and presents a case study on the importance of a balanced logistical organization from the 17th century, after which formal definitions and objectives are introduced. In addition, we discuss relations with other management areas as well as with other science domains such as law or social and political sciences. Topics and concepts in this chapter are discussed at an elementary level, aiming to provide an introduction to the topical field of operations, logistics, and supply chain management.

3.1 History From the first societies and economies in Egypt, Mesopotamia/Persia, China and Central America onwards, the wellbeing of mankind depended on the ability to transport and store goods effectively and efficiently. With the innovations of the wheel and the ship and their widespread use, the wealth of societies increased manifold due H. Zijm (B) Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, Overijssel, The Netherlands e-mail: [email protected] M. Klumpp FOM University of Applied Sciences and University of Duisburg-Essen, Essen, Nordrhein-Westfalen, Germany e-mail: [email protected] S. Heragu Industrial Engineering and Management, Oklahoma State University, Stillwater, OK, USA e-mail: [email protected] A. Regattieri Department of Industrial Engineering, University of Bologna, Bologna, Italy e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_3

27

28

H. Zijm et al.

to efficient production, centralized storage as well as trade with other regions and the introduction to exotic products. Well-known monuments as the Egyptian pyramids, the enormous temple and living quarters of civilizations in Central and South America as well as in Asia testify to the transportation and logistics management capabilities of the respective societies and economies. In Roman times, the establishment of the road system and the post system based on this as well as the Roman ship fleet enabled their rulers to build one of the first “global” empires. In addition, the planning and implementation of the water transport system via long-distance viaducts was one of the great achievements, which enabled urban living and culture in a way never seen before in history. In another part of the world, the destruction of the great Chinese fleet in the 15th century marked the beginning of a century-long downturn of Chinese economic, trade and military power in the region. It was in this time that European countries started to support long-sea travel enterprises and word-wide trading, thereby changing the face of the global economic and political map, as evidenced for example by the constitution of the Dutch East Indies Company in Amsterdam, which formed the basis of Dutch power in the seventeenth century. Later, the industrial revolutions marked the supremacy of innovation-based industries in western countries. In the last four decades we have witnessed a turn again as Asian countries, in particular Japan, developed highly competitive manufacturing systems based upon quality improvement and waste reduction philosophies, for example in the automotive and electronics sector. The massive industrialization taking place in countries like China and Korea, with initially very low wage rates, also helped to develop those countries into formidable economic powers. Additional huge investments in sea shipping and port development were instrumental in establishing their solid position in today’s global trade: the four largest cargo harbors of the world are currently all Asian. It is well known that the development of transportation and logistics has often been intertwined with military operations. The Punic wars between Rome and Carthage with the famous March to Rome via the Alps by Hannibal are tales and concepts still taught today in military academies. They also are benchmarking endeavors for logistics management. Military operations have continued to influence logistics: the large building projects in railroad infrastructure of the 19th century in the US, Germany and Russia as well as other European countries served as “dual use” logistics backbones and indeed also the two world wars of the 20th century have accelerated the development of logistics as a science. From modern developments such as the forklift and transport boxes (containers) to the establishment of secure communications (internet)—in many cases military and general economic objectives in transportation and communication went hand in hand. In the first acknowledged economics book by the philosopher Adam Smith in 1776, logistics and transportation was named as an important enabler. In his “Wealth of Nations” he describes how long-distance transport enables market extensions and therefore helps to bear the full fruit of the principle of division of labor:

3 Operations, Logistics and Supply Chain Management …

29

As by means of water-carriage a more extensive market is opened to every sort of industry than what land-carriage alone can afford it, so it is upon the sea-coast, and along the banks of navigable rivers, that industry of every kind naturally begins to subdivide and improve itself.1

Scientists and industrial engineers “avant la lettre” like Charles Babbage, Frederick Winslow Taylor and Frank and Lilian Gilbreth, helped to develop the division of labor principle into a full-blown theory of economic manufacturing, practiced among others by Andrew Carnegie and Henry Ford (Chandler 1977). Efficient mass production continued to be the dominant philosophy until the early sixties of the preceding century, when product diversification and mass customization started to change the landscape. Faced with the development of global markets, multinational organizations were to manage complex supply chains consisting of many independent companies (raw material suppliers, component manufacturers, OEM’s, logistic service providers, etc.). If not properly coordinated, such supply chains may grow untenable, with very long lead times, high stock levels and still at best moderate customer service. Together with marketing management, logistics and supply chain management has since then emerged into the modern form of business and management science we see today, to provide an answer to the demands and realities of modern global economies.

3.2 Case Study: The VOC and the Birth of the First Worldwide Supply Chain 3.2.1 Historical and Geographical Background of the Dutch East Indies Company Already in the 15th century the Dutch Republic (officially known as the Republic of the Seven United Netherlands) held important positions as a trade nation, primarily in bulk products. Due to for instance logistical innovations as the Flute, a relatively cheap ship designed and built in the Netherlands and able to transport large amounts of cargo, the Dutch dominated the trade in bulk goods such as grain and wood from the Baltic states. In the 16th century, the Portuguese had explored sea routes to the East, the knowledge of which was made public by the Dutch cartographer Plancius. Portugal initially controlled the supply of pepper, subsequently traded in Amsterdam by a syndicate of European trading houses. Unfortunately, the Portuguese were unable to secure the pepper supply due to (English) piracy, transport problems and capacity shortages, so the Dutch decided to take control of the supply themselves. Several Dutch cities, in these days still relatively

1 Cf.

Smith (1776—reprinted in 1994), p. 20. Interestingly, the recent economic development of China along its sea-coast shows that this quote is as actual as ever.

30

H. Zijm et al.

independent, sent out their own fleet which returned successfully. Recognizing that pooling of financial and maritime resources might help to better withstand the threats of piracy, the then Secretary General of the Republic, Johan van Oldebarneveldt, managed to bring together the cities of Amsterdam, Middelburg, Rotterdam, Delft, Hoorn and Enkhuizen, leading to the birth of the Dutch East Indies Company (Verenigde Oost-Indische Compagnie, or VOC) in 1602. From the very beginning, the management of the VOC was in the hands of an executive body (the Heeren XVII), representing the various Chambers (the six constituting cities). Later, the inception of the West Indies Company (WIC) organized trade with parts of Africa and America in the same way. The expansion of trade to overseas areas in Asia, Africa and America of what became known as rich trade, in e.g. pepper, sugar and especially spices, yielding considerable larger margins than bulk trade. The latter already had provided the basis for a well-developed shipbuilding industry, including many suppliers, while as a seafaring nation it was not hard for trade companies to hire crew. The economic interest in luxury goods grew rapidly: in the middle of the 17th century the value of bulk trade amounted to 3 million guilders while the rich trade (luxury goods) already reached a value of 20 million guilders. Interestingly, the Dutch quickly discovered that the manipulation of markets through strategic stock keeping might lead to considerable additional profit. A sort of market hierarchy was developing in Europe that resulted in the balancing of supply and demand through an inventory or ‘staple’ market, where traders stabilized the prices with a dynamic stock keeping policy. Market manipulation became an established business practice. It is this rich trade that eventually provided the basis for the wealth of the Dutch Republic in the 17th century and the position of Amsterdam as the world’s financial center, giving impulses also to arts and sciences (Simon Stevin, Rembrandt van Rijn, Christiaan Huygens, Baruch Spinoza). The VOC and the WIC built trade posts all over the world that became part of a complex supply chain with Amsterdam, and to a lesser extent Middelburg, as the European centers. In the 17th century, the VOC had 20 settlements in Asia, with 11.000 employees while in the Netherlands 3000 people were directly employed and many thousands indirectly via suppliers; it is therefore often viewed as the first multinational company ever. In what follows, we concentrate on the logistics network built by the VOC.

3.2.2 Batavia as the Center of the VOC Logistical Network in the East The primary reason for the inception of the VOC had been the desire to set up a secure supply of pepper and spices to the Netherlands. These spices could

3 Operations, Logistics and Supply Chain Management …

be found mainly on the Indonesian islands (the Spice Islands), with eventually Jayakarta, later named Batavia (and currently known as Jakarta), as the center of Asian trade. It was certainly not self-evident that Jayakarta would become the headquarters of the VOC from which all activities would be coordinated. The focal point of the Dutch activities in Asia for the early period was the Spice Islands, but this location was impractical from a purely nautical point of view. Due to the monsoonal cycle, the Spice Islands were only accessible for six months a year while return shipping was only possible in the other half of the year. Early in the 17th century, the need for a warehouse arose, i.e. a place to store goods to be reloaded onto other ships for which a yard and building in Bantam (NorthWest Java) was allowed. The problem however was that Bantam charged high import and export taxes. A location with the same logistical advantages but independent from the local authorities was obviously more suitable. Eventually, Jayakarta was selected to become the administrative and logistics center. The bay of Jayakarta with the ‘thousand islands’ was ideally suited for both shipbuilding and repair. In 1619, Jayakarta was renamed Batavia by Jan Pieterz. Coen, and after 1620 it became the logistic headquarters of the VOC. From that time on, all major decisions relating to administration and logistics were made there. The stranglehold that VOC Batavia had on the Asian shipping network became even stronger after the directors in the Netherlands decided that all return shipping should be routed through Batavia. Initially, some retour ships were still sailing directly back to Europe from the Coromandel Coast, Surat or Persia. This was eventually forbidden by the VOC in the Netherlands mainly for safety reasons, but this centralizing policy was very much welcomed by VOC Batavia in order to reinforce their central role. The management in Batavia was, in principal, bound to the directions they received from the Netherlands for the return cargo. However, to a certain extent Batavia was able to influence the policy by their implementation of the return shipping. To understand the difficulties in synchronizing the Europe-Asian and IntraAsian networks, let us consider the main sailing schedules in some more detail. Ships usually left the Dutch Republic in two main periods: Christmas and Easter, and arrived in Batavia between June and October. The departure date of the retourships was dictated by the VOC in the Dutch Republic and based on the desire for the fleet to return to Europe before or at the start of the European autumn. Any later arrival would prevent further European distribution to the buyers of the VOC products because of autumn storms and northern winter ice. This timing however, created problems for the VOC in Asia. To arrive at this time, the fleet had to leave Asia around the turn of the year, meaning all supplies from other parts of East-Asia should have arrived in time. A complicating factor was the fact that additional to the premium products ballast was required for the specific design of the Dutch vessels, for reasons of

31

32

H. Zijm et al.

stability. Ballast loading formed an important logistical aspect of the organization. Therefore, merchants within the intra-Asian network were permanently looking for suitable goods that could be used as ballast, in particular to find ‘paying’ ballast that would contribute to the profit rather than being an expense. The choice of ballast played an important role in the efficient loading of ships, ensuing that the cargo capacity was optimally utilized by valuable products. The most lucrative ballast appeared to be sugar from Taiwan and copper from Japan but ships from this region had problems reaching Batavia in time as they could generally only sail from the Far East after September. A partial solution for this problem was found in sending the early arrival retourships to Taiwan or even Japan to collect their return ‘ballast’ themselves. They could keep part of their European cargo specially shipped for Japan on board and could supplement that with products like pepper and sandalwood from stocks in Batavia, and silk from Persia, brought to Batavia by other VOC ships, and also traded via the Dutch trading post on the Japanese island of Decima. These ships would return to Batavia in November-December with the bulk of their European cargo in their hold and only needed to be ‘topped up’. The timing of shipping was less critical for the important spices from the eastern region. There was a guaranteed supply of these spices and purchasing costs were low, both of which allowed the VOC to keep excess stocks without putting a financial burden on the VOC. The VOC made optimal use of their ships by sending them at the end of the western monsoon in March, so they could return with the change of the monsoon in April. The nature of the organization in Taiwan was determined by local (indirect) trade with China and the strictly regulated trade with Japan on Decima. The Chinese goods assembled in Taiwan had to be sent to Japan in a short time frame due to the strict dates that were set by the Japanese authorities and the monsoon conditions. Subsequently, the goods returning from Japan had to be sent off as soon as possible to Batavia together with the Chinese goods for the return fleet or to the retourships in the Pescadores that had sailed on from Batavia to the Far East. In the Middle East, the ships that ended up in Persia could make their return voyage to Batavia directly from around May but often they sailed via Galle at Ceylon to make a stopover for horses that were traditionally shipped from this region. From here, an extra trip into the Bay of Bengal was also possible in order to spend the money received from trade in Persia. However, the VOC also had to work around the necessity for the Persian silk to be in Batavia in time for transport to the Far East or with the return fleet. In conclusion, the VOC demonstrated a remarkable logistics organization in optimizing their shipping around the arriving and departing fleet for Europe, despite the many meteorological and political complications. Through flexible employment of a differentiated fleet, they were able to attune and synchronize

3 Operations, Logistics and Supply Chain Management …

33

as much as possible the needs of both the European-Asian and the Intra-Asian network. In the 18th century businesses gradually turned downwards. The Dutch lost their position as a maritime power mostly to England; after the fourth EnglishDutch war retourships were even no longer able to reach the Republic. At the end of that century, only trading posts at Decima, in Kanton and in Java were still in the hands of the Dutch. In 1799 the closure of the VOC was definitive.

3.3 Definitions There exist several definitions for supply chains, logistics and their constituting operations. For this textbook, we follow the definitions provided by the American Council for Supply Chain Management Professionals (CSCMP), albeit the wording has been slightly adapted to include reverse flows and to emphasize the distributed nature of modern supply chains. A supply chain encompasses all activities (operations) needed to convert raw materials into final products, from sourcing through component manufacturing and final assembly to distribution to end-markets, and including all necessary materials handling and storage (in short, logistics) activities. More and more, it also includes the handling of return flows of products and possible re-use of materials and components, in which case we speak of closed loop supply chains. Almost always, these activities are not executed by only one industry but instead encompass a number of companies and organizations jointly operating in a chain or network, referred to as end-to-end supply chains. Logistics refers to the transportation and storage of materials, parts and products in a supply chain. Logistics includes inbound and outbound processes to and from warehouses, as well as internal and external materials handling and transport operations. It also includes the execution of services and the transfer of information between the various stages of a supply chain. Once these core topics have been defined, one might easily state that Supply Chain Management and Logistics Management is the management of supply chains and logistics, respectively. Although clearly true, some further clarification may be helpful. Therefore, we have chosen to provide separate definitions. Supply chain management encompasses the planning and management of all supply chain operations. Importantly, it also includes the coordination and collaboration with channel partners, which can be suppliers, intermediaries, third party service providers, and customers. In essence, supply chain management integrates supply and demand management within and across companies. Logistics management is that part of supply chain management that plans, implements, and controls the efficient, effective forward and reverse flow and storage of

34 Fig. 3.1 Hierarchy of supply chains, logistics and transportation

H. Zijm et al.

Global Supply Chain Management

Logistics Management

Transport Operations

goods, services and related information between the point of origin and the point of consumption in order to meet customers’ requirements. It typically includes inbound and outbound transportation management, fleet management, warehousing, materials handling, order fulfillment, logistics network design, inventory management, supply/demand planning, and management of third party logistics services providers. A proper execution of logistic operations depends on, and has impact on, sourcing and procurement, production planning and scheduling, packaging and assembly, and customer service. Logistics management is an integrating function, strongly relying on an adequate information infrastructure, and ideally synchronized with other functions including marketing, sales, manufacturing and finance. Unfortunately, in many popular magazines and in the eyes of the public, logistics is often still seen as synonymous with transport. The above definitions clearly indicate the much wider scope of logistics but to avoid any ambiguity, Fig. 3.1 once more relates Supply Chain Management, Logistics Management and Transport Operations to each other.

3.4 Key Aspects of Supply Chains and Logistics The definitions presented above deserve some further discussion to put them in perspective, even more so when they are projected on current societal needs and concerns. Below, we elaborate on these definitions by discussing seven key aspects. Availability of materials, products and information This is the basic notion from which the term “supply chain” stems. Its primary objective is to ensure the timely availability of the right quantities of raw materials, parts and products together with all information needed for further processing at their destined locations. Functions in a supply chain that are instrumental in achieving this goal are sourcing and procurement, transport and logistics, manufacturing and assembly, stock keeping between subsequent phases, and sales. With each step, data on preceding steps and product characteristics is used, and information for subsequent

3 Operations, Logistics and Supply Chain Management …

35

Fig. 3.2 SCOR model of end-to-end supply chains (Poluha 2007)

steps is produced. Note that a product may also encompass a service delivered, often realized by means of physical products as a platform. It is important to realize that usually supply chains encompass various companies; for each company, its customer may be the next stakeholder in the supply chain (see also Fig. 3.2). Coordination of the goods flow throughout the supply chain hence requires the synchronization of activities that are controlled by different stakeholders, each with their own objectives. This makes overall supply chain synchronization a formidable task. Cost-efficiency For a long time, cost-efficiency has been the dominant measure of performance of production and logistic processes, meaning that a maximum output should be achieved with minimal materials and manpower utilization. For that reason, initial phases of industrial production were typically characterized by large batch manufacturing and low product variety, and machines and operators were considered to be cost factors. The division of complex production processes in a large number of simple repetitive tasks, each to be performed by one operator or man/machine combination such that learning curves are optimally exploited, has long been the management principle that characterized most mass production facilities. Gradually however, it became clear that cost-driven production management may lead to high levels of efficiency at individual stages in a supply chain, but not necessarily to minimal costs for the overall chain. For instance, large production batches as a result of efficient production by avoiding costly changeovers lead to high work-in-process stocks between subsequent stages and hence to high work-in-process inventory costs. As long as product variety is low and the same or similar products are produced for many years, at least these stocked products are ultimately sold and add to overall revenue (although after a considerable time elapse). But if demand variety increases and product life

36

H. Zijm et al.

cycles become shorter, companies even run the risk of obsolescence, i.e. products in stock that are no longer demanded and have to be dumped. The implications of costefficiency as a dominant performance index therefore change fundamentally when moving from an individual to a system level. That observation eventually led to the development of lean management principles, see below under “Effectiveness”. Customer orientation From the sixties of the preceding century onwards, consumer markets started to change. No longer products were accepted as they were; customers demanded more product variety and companies started to realize that they might distinguish themselves from competitors by responding to these demands. The push market gradually changed into a pull market, characterized by customer differentiation, shorter product life cycles and tailored solutions. More attention was given to product quality, again recognized as a distinguishing feature and indeed an important factor in gaining customer trust and loyalty. The basis of almost any marketing campaign today is the recognition of customer profiles which in turn provides the foundation for market segmentation. In particular, e-commerce offers ample opportunity to develop and refine customer profiles, and to use them to offer tailor-made products or services. Within production management, the classical distinction between mass production, small batch manufacturing and one-of-a-kind production (cf. Hayes and Wheelwright 1979) remains valid but the current opportunities to track customer wishes, together with high levels of automation and the resulting flexibility, offers unprecedented opportunities for mass customization. Speed Within a customer-dominated market, timeliness of delivery quickly became a key requisite in supply chain management. The classical way to achieve timely delivery was to keep sufficient anticipation stocks at a large number of locations, certainly for those multinationals that were selling all over the globe. High inventories however represent large amounts of capital invested, hence limiting liquidity of companies involved. High interest rates, as they occurred after the two oil crises of 1973 and 1979, render such stocks untenable, while moreover long cycle times make it impossible to quickly react to rapidly changing market demands. Already in the early sixties, Forrester (1961) explained how final demand variation may be amplified upstream, with large stocks and inflexibility as a result, a phenomenon that later became known as the bullwhip effect (Lee et al. 1997). Short lead times and supply chain synchronization across the various stakeholders in principle offer an adequate remedy, as expressed by the phrase “know your customer’s customer”. Winners are those companies that are able to flexibly produce a large variety of products with high speed, optimally responding to observed or forecasted market demand. Effectiveness Lean production and lean logistics refer to the ability to limit activities in a supply chain to those that really add value to products and services as experienced by a customer, and that avoids any activity that can be considered as waste (e.g. Womack et al. 1990). An early implementation of Lean Production was the Just-In-Time

3 Operations, Logistics and Supply Chain Management …

37

management philosophy originating in Japan which might be viewed as a first radical attempt to eliminate any unnecessary stock by fundamentally rethinking the underlying production processes. If large set-up times between production of different product variants cause large batch production to remain efficient, but lead to unwanted high stocks and long lead times, then start to reduce these set-up times instead of accepting these large batches (Shingo 1985). More fundamental, many supply chains were characterized by “frozen” technological capabilities whereas the Japanese philosophy basically pointed to the fact that no current practice needs to be accepted as “given”. In a similar way, quality assurance has been severely influenced by a switch from product quality control to process quality control, again inspired by the notion that producing malfunctioning products is simply a waste of time and effort and should be avoided. Environmental Sustainability In the last two decades, it has become apparent that current production and logistics systems cause serious and in the long run unacceptable environmental damage, due to for instance the emission of hazardous materials (CO2 , NOx , and particulate matter), congestion, stench, noise and more general the high price that has to be paid in terms of infrastructural load. In addition, some natural resources become scarce and also are not evenly distributed in terms of type and geographical location in the world. Logistic chains enable the distribution of materials, food and products from the locations where they are extracted, harvested or produced to people’s homes and nearby stores. Current supply chains and logistics systems are global, partly due to natural conditions but certainly also because of labor rate differences between emerging and mature economies. First indications of reshoring production however become visible, not only because wages are moving upwards in a number of Far-East countries, but also since the amount of manual labor needed in high tech products continues to diminish, while logistic costs are increasing. As a result, future supply chains are believed to be “glocal”: global when needed, local when possible. On the other hand, global supply chains will remain indispensable in cases where conditions for growing food ingredients are only satisfied in some regions in the world, or when essential minerals are only locally available. Social aspects There is growing attention for the need to create safe and socially acceptable working conditions, but so far these conditions have not been realized everywhere, certainly not in a number of developing countries. In addition, ageing continues and although official retirement ages are expected to further increase, there nevertheless will be a balance change between workers and retired people, in favor of the latter. To respond to these challenges and to keep pensions payable, productivity has to rise while at the same time diminishing the ecological and social footprint. This undoubtedly requires a quality upgrade of the human resource pool, e.g. by better education and training, including lifelong learning programs. In parallel productivity can be improved by better support tools, easier access to relevant information, and last but not least rapid automation of both technical processes (robotics, automatic vehicles) and decision

38

H. Zijm et al.

making (i.e. artificial intelligence), to diminish the need for human workers to execute routine tasks. We believe two technologies in particular—autonomous vehicles and virtual reality—will significantly impact many aspects of our life, including operations, logistics, and supply chains. Autonomous vehicles including unmanned aerial vehicles (drones) and driverless vehicles (cars and trucks) will help the quest for mass customization and facilitate the growth of a sharing economy. It is not too far-fetched to imagine a drone routinely making a parcel or pizza delivery on demand. Driverless cars in a city’s downtown area will provide personalized transportation. Both driverless cars and trucks can be operated such that the capacity is better utilized in a sharing economy. Virtual reality already has the capability of providing realistic sensations (touch, sound, and sight) so that users can visualize the real world by operating in a simulated, three-dimensional environment. There are many aspects of operations and logistics that can be augmented, enhanced or even replaced by virtual reality—training of personnel including technical and medical professionals, collaborative design of products, marketing of retail products including clothing, design of buildings, etc. Additionally, for the sake of completeness, we mention the 7R objective, as it is often viewed as a good summary for the main objective of logistics. Logistics should provide: • • • • • • •

The Right goods (i.e. as ordered) In the Right quantity With the Right quality At the Right time At the Right place At the Right cost At the Right sustainable impact/footprint.

3.5 Relations to Other Scientific Disciplines In order to fully understand the impact of operations, logistics and supply chain management a discussion of relations with other science fields and objectives is in place. Without claiming to be complete, we mention the following. • Technology and Engineering: Many technological advancements in transport technology but also in manufacturing pave the road for more efficient and effective supply chains, with less impact on the environment. The design of cleaner engines and the use of alternative fuel and energy sources for transport means are notable examples. In particular, the use of LNG (Liquid Natural Gas) as fuel for trucks and the application of electric cars in city distribution reduce carbon emission substantially. Also, the application of additive manufacturing (better known as 3Dprinting) in parts production not only reduces the use of raw materials but has also important logistic consequences, in particular in slow moving spare parts supply,

3 Operations, Logistics and Supply Chain Management …

39

since it allows for the manufacturing of parts where and when needed, hence reducing anticipating safety stocks. Finally, we mention the already well-established application of automated materials handling devices (Automated Guided Vehicles, unmanned storage and retrieval equipment) as well as robotic manufacturing and assembly. Currently, first trials take place with unmanned road transportation and platooning; in addition, drones and larger unmanned cargo aircraft are expected to deliver packages at remote locations not easily reachable otherwise. With respect to surface transport, much is expected from the further application of MAGLEV (short for magnetic levitation) trains that float over a guideway (often a monorail) using magnetic power instead of a wheel base and seem to be a viable alternative for cargo transport on medium range distances. • Sociology: Since 2007, more than half of the world’s population is living in urbanized areas, in western countries, city inhabitants constitute already more than 70% of their total population and the numbers are still increasing. Cities therefore are centers of economic and cultural activities and naturally require a superb logistic infrastructure to take care of all forward, inner city and reverse flows. Unfortunately, in particular older cities were not built to deal with such intensified traffic and hence face enormous infrastructure construction challenges. As mentioned earlier, the shift in balance between workers and retired people (even taking into account the increasing retirement age) also constitutes serious problems in a number of sectors, including logistics. The necessary automation to achieve a further productivity rise will also change the nature of jobs, from operational to supervisory, and hence requires significant additional training at all levels. Next to this comes the rapid adaptation of social media as an easy communication means, leading for instance to a further increase of e-commerce business. Delivery-at-home next day has already become the norm but can only be reached with a superbly functioning logistics system. • Law and Politics: Global logistics is by definition border-crossing logistics which requires intrusive inspections by e.g. Food and Drug Administration authorities as well as confirmation to customs and tax regulations. Such inspections require intensive document handling, not in the least place to prevent illicit trade of e.g. drugs and even human trafficking, but also tax fraud, intellectual property rights violation and counterfeiting. Unfortunately, laws and legal regulations in many European countries are still not harmonized, therefore prohibiting a smooth synchronization of (electronic) documents. But also at a national or even regional level, legislation is often far behind when it comes to experiments with technical innovations. One example concerns the transition to multimodal transport, requiring the availability of multiple modes such as inland water vessels, next to road and rail infrastructure and the availability of terminals that serve as cross-docking facilities from one mode to another. All these facilities place a heavy demand on the public infrastructure and are therefore often subject to long and cumbersome negotiation and decision processes that involve many stakeholders. When it comes to the introduction of unmanned transport, again major legal obstacles have to be dealt with, in particular accountability issues in case accidents happen. As in other

40

H. Zijm et al.

disciplines, such debates strongly hamper the introduction of new and innovative techniques, also in logistics and supply chain management. • Informatics and Artificial Intelligence: A prerequisite to any modern manufacturing and supply chain planning system is the ability to process large amounts of data. This was the success factors of early materials requirements planning (MRP) and enterprise resource planning (ERP) systems. These systems however turned out to be primarily administrative rather than smart planning systems; the latter require sophisticated Operations Research techniques to integrate in particular materials, capacity and financial planning and control. Integration with demand market forecasting, based on e.g. customer profiling, and on the other hand supplier performance rating may lead to an explosion of data needed, not in the least place since the rapid advance of e-commerce, and social media yield a wealth of data useful for marketing purposes. Next to classical statistical analysis, new data and process mining techniques have proven their value in translating these data into useful information, which in turn yields input for supply chain planning. Smart planning based on a thorough data analysis has received considerable attention, again due to the computer power available nowadays in combination with a wealth of optimization techniques that have been developed. Without claiming to be complete, we mention advances in mathematical programming and combinatorial optimization (including a large variety of sophisticated heuristic techniques), stochastic systems analysis, multi-agent systems, cooperative and non-cooperative game theory, input-output analysis and artificial intelligence based systems. Finally, we should mention the advance of cloud computing and cloud logistics as an important recent development. • Biology and Environmental Sciences: The living environment of our planet is a major concern also to operations, logistics and supply chain management. To date, emission of carbon and particulate matter due to manufacturing, warehousing and transport operations is still rising, both absolutely and relatively. To diminish the footprint of transport in particular, possibilities for horizontal and vertical cooperation in supply chains are explored, with special attention for supporting business models. Switching of transport from road to rail or water is not only more efficient but also a way to diminish congestion as well as hazardous materials emission. In addition, there is increasing interest for the re-use of products or their constituting parts and components in a second life (closed loop supply chains), thereby diminishing the number of products ending at a garbage dump, or even by harvesting basic materials and use them as a source for new production processes (cradle-tocradle or circular economy). Such return flows however are not easy to manage, partly because the availability of return products is subject to uncertainty, in terms of both time, quantity and quality. Interestingly, we also observe a growing interest in many disciplines for bionic solutions (the nature as blueprint for intelligent and cradle-to-cradle concepts) and the chance to learn from environmentally balanced systems in terms of closed-loop dynamics.

3 Operations, Logistics and Supply Chain Management …

41

3.6 Further Reading and Links One of the earliest sources is the famous book “The wealth of nations”, written by Adam Smith and initially published in 1776, in which he among many other things discussed the efficiency gain that can be achieved by means of specialization, see the reprinted version that appeared in 1994. Smith’s work on labor division was further explored by Babbage (1835) who later became known as the father of the principles of the digital computer, while a very thorough introduction into mass production and distribution in the US is provided by Chandler (1977). Klaus and Müller (2012) have composed a well-known reader of classical contributions to the history and conceptual foundations of Logistics as a scientific discipline. Another interesting source on military logistics in particular is Van Creveld (2004). The case study on the Dutch East Indies company draws heavily on the dissertation of Parthesius (2010) while furthermore the book of Gaastra (2002), published in Dutch, appeared to be a rich source. With respect to the way technological progress may influence our economy and society in general, a plethora of views has been published. Rifkin (2014) predicts that the advance of the Internet of Things eventually leads to a situation in which the marginal cost of producing and sharing a wide range of products and services becomes almost negligible, just like many information goods. A discussion on the potential of artificial intelligence and its social consequences is provided by Bostrom (2014). Broad introductions towards societal implications, in particular inequality and remedies against it, are provided by Sachs (2005) and Acemoglu and Robinson (2012).

References Acemoglu D, Robinson JA (2012) Why nations fail—the origins of power, prosperity and poverty. Crown Business/Random House, New York Bostrom N (2014) Superintelligence. Oxford University Press Babbage CH (1835) On the economy of machinery and manufactures, 4th edn. Charles Knight, London Chandler AD (1977) The visible hand—the managerial revolution in American business. Belknap Press/Harvard University Press, Cambridge, MA Forrester JW (1961) Industrial dynamics. MIT Press, Cambridge, MA Gaastra FS (2002) The history of the VOC (De Geschiedenis van de VOC, in Dutch). Walburg Press, Zutphen, the Netherlands Hayes R, Wheelwright S (1979) Link Manufacturing process and product life cycles. Harvard Business Review, pp 133–140 Klaus P, Müller S (2012) The roots of logistics—a reader of classical contributions to the history and conceptual foundations of the science of logistics. Springer, New York Lee H, Padmanabhan P, Whang S (1997) Information distortion in a supply chain: the bullwhip effect. Manag Sci 43(4):546–558

42

H. Zijm et al.

Parthesius R (2010) Dutch ships in tropical waters—the development of the Dutch East India Company (VOC) shipping network in Asia 1595–1660. Amsterdam University Press, Amsterdam, The Netherlands Poluha RG (2007) Application of the SCOR model in supply chain management. Youngstown, New York Rifkin J (2014) The zero marginal cost society. Palgrave Macmillan Sachs JD (2005) The end of poverty. Penguin Shingo S (1985) A revolution in manufacturing—the SMED system. Productvity Press, Portland Smith A (1776, reprinted in 1994) An inquiry into the nature and causes of the wealth of nations, Random House, New York Van Creveld M (2004) Supplying war: logistics from Wallenstein to Patton. Cambridge University Press, Cambridge Womack JP, Jones DT, Roos D (1990) The machine that changed the world. Maxwell Macmillan, New York

Part II

Key Domains of Supply Chains

Chapter 4

Purchasing and Supply Management Holger Schiele

Abstract Purchasing is the function in a firm responsible for the professional management of a firm’s interface with the supply market, to ensure its supply with the necessary goods and services provided by other organisations, i.e. suppliers. Industrial firms spend more than half of their turnover on supplies, which is why the purchasing function has become a central success factor for modern firms. Purchasing can be distinguished into strategic sourcing (supply planning, supplier selection and contracting) and operative procurement (material ordering, expediting and paying). The activities of a purchasing department can be organised in a purchasing year cycle, which repeats on an annual basis. Next to ensuring a safe and timely supply, purchasing has the target to achieve good costs as well as to contribute to innovation and improve the strategic position of a firm. For that several tools have been developed, such as the Kraljiˆc-Matrix (which helps to develop sourcing tactics), the lever analysis (used to systematically achieve cost savings) or the preferred customer approach (used to achieve competitive advantages through smart purchasing).

4.1 History and Relevance (Basic) The installation of the inter-state railways in the U.S. has provided much impulse—not only for the development of the management function in general but also for the development of the purchasing function. For instance, at the Pennsylvania Railroad Company purchasing was given departmental status as early as in 1866. Also, the allegedly first management book in the English language that was fully dedicated to purchasing was published a few years later by a railway executive, Marshall Kirkman: “The handling of railway supplies: their purchase and disposition” (1887). Concerning the establishment of a dedicated purchasing department already by 1910 Redtmann noted in a dedicated journal article: “It should be regarded as a big mistake to neglect the significant advantages of a well-organized purchasing H. Schiele (B) University of Twente, Enschede, The Netherlands e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_4

45

46

H. Schiele 100

Disposal of revenue (%)

Profit 80

Wages, salaries and overheads

60 40 Materials and services

20 0 1940

1950

1960

1970

1980

1990

today

year

Fig. 4.1 Growing importance of purchasing (Baily et al. 2008, p. 12)

department. Experience shows that this negatively impacts organisational success. Every experienced businessman will establish a purchasing department.” (Redtmann 1910, p. 55). Despite these and sporadic other beginnings every few years, though, the purchasing function largely remained in a state of the “sleeping beauty” that was only really awakened in the 1990s. Today, an industrial company on average spends about 60% of its turnover on supplies (see Fig. 4.1). A few decades ago, this was completely different. For instance, in its early days Volkswagen ran its own farm that grew pigs, then produced the sausages, and then served those sausages as lunch for the workers. Today, catering is usually outsourced to firms concentrated on the core competence of running canteens, who again rely on specialists for diverse types of food and so on. The early VW Beetle purchased no more than 15% of its supplies. The remaining value was added by the VW factory. Porsche Cayenne is the exact opposite. This contemporary car relies on 85% of purchased components, and only the smaller fraction is value added by the manufacturer. Similar to the automotive industry, the depth of production has decreased substantially in virtually all industries; manufacturing leads the way, and service companies follow. Because of the growing importance of a well-managed purchasing function, it can be derived that “Firms exist by selling, but earn profits by buying.” This statement becomes clear by a simple calculation. Assuming a typical industrial firm reduces its purchasing costs by 10%, then it would double its return on assets. Figure 4.2 illustrates how savings in purchasing lead to improvement in a company’s bottomline. It has been found that more mature purchasing organisations achieve significantly higher savings and that the relation maturity—savings can be measured (Schiele 2007). This means that from a firm’s perspective, it is worth to invest in increasing the sophistication of the purchasing department (i.e., to have better planning tools, well-defined processes, an adequate organisational structure, differentiated roles,

4 Purchasing and Supply Management Sales $1 Million *

Inventory $150.000 ($135.000) Sales $1 Million

Minus † Total Cost $950.000 ††($900.000)

Total Assets $500.000 ($485.000)

47

Divided by

Investment Turnover 2 (2.06) Multiplied by

ROA 10% (20,6%)

Profit $50.000 ($100.000)

Divided by

Sales $1 Million

Profit Margin 5% (10%)

* Inventory is approximately 30 percent of total assets. † Purchases account for half of total sales, or $500.000. †† Figures in parantheses assume a 10% reduction in purchase costs.

Fig. 4.2 Small reductions purchasing volume substantially increase bottom-line results (Johnson et al. 2011)

well-educated purchasers) and rely on an efficient controlling system enabling continuous improvement.

4.2 Definition and Objectives (Basic) There are several definitions of purchasing, which have been used somewhat differently in Europe and in North America. To make things more complicated, these definitions have also changed over time. A development of particular importance in this context is to split the purchasing function into a more strategically oriented set of activities around selecting and contracting suppliers and more operatively oriented activities ensuring the ordering and delivery of materials and services. The reason for this split is twofold: on the one hand, an often observed phenomenon is that if strategic and operative activities are bundled in one job, the preponderance of operative day-to-day activities prevent the execution of more strategic and long-term activities. On the other hand, strategic and operative activities require different skills and differently educated people. This is illustrated in Fig. 4.3. Strategic sourcing comprises the process of planning supply, selecting suppliers and contracting them in order to establish the potential for supply. Operative procurement encompasses the ordering of material and services, ensuring its delivery and, finally, activating the payment, thus executing the order. Strategic sourcing and operative procurement together represent purchasing. Thus, a purchasing department is responsible for both operative and strategic activities. Purchasing (or supply man-

48

H. Schiele Purchasing

Planning supply

Selecting supplier

Contracting supplier

Strategic Sourcing

Ordering material

Expediting order

Paying delivery

Operative Procurement

Fig. 4.3 Strategic sourcing and operative procurement

agement) is the strategic and operative process of supplying an organisation with materials and services from sources external to that organisation; the purchasing department is active in all situations, which require a payment to third parties. In principle, it is possible to run the strategic sourcing process without following it by operative procurement activities, for instance, if the demand for a good unexpectedly disappears but a supplier had previously been selected. At the same time, in some cases, it is also possible to execute operative procurement activities without relying on strategic sourcing activities. Ordering material without relying on a contract or, more generally, without following the established corporate process is called maverick buying. Maverick buying is considered to be detrimental to purchasing success, because uncontrolled purchases cannot professionally be handled. For instance, volume discounts are made impossible if users order material independently from each other and independent from established processes. In order to make it possible for an organisation to maximally leverage its purchasing power, all activities that result in an invoice from an external organisation should be managed by a professional purchasing department, hence: no payment without a purchase order. Traditionally, purchasing has three main objectives: (1) ensuring safe, timely and sufficient supply at (2) appropriate quality with (3) the lowest possible costs. Reflecting the growing importance of suppliers for the prosperity of a firm, two novel objectives may be added, namely, (4) facilitating innovations from and with suppliers and (5) ensuring competitive advantage to the firm by guaranteeing privileged access to sources of supply. 1. Safe supply: the most basic objective of purchasing is to provide the materials or services needed to execute the transformation process of the respective organisation (i.e., the right goods need to be at the right quantity at the right time and, of course, at the right place, if needed with the required flexibility to adjust to changes). Stopping an assembly line, for instance, causes substantial costs that far exceed the value of missing components; hence, the emphasis on this objective is a necessary condition.

4 Purchasing and Supply Management

49

2. Quality: is another necessary condition, because a product that does not match required quality criteria cannot be sold. Sustainability of the supply chain has recently been included as a special and distinguished manifestation of quality. 3. Cost: is traditionally the main sufficient condition to make a sourcing project feasible. Regarding the importance of the cost block “supplies” in a modern firm, this criterion gains in importance. It is worth noting that costs exceed the price of a component but include logistics costs (transport, handling, storage) and several other types of costs including costs of utilisation, maintenance and extending to recycling costs (total costs of ownership) as well. 4. Innovation: Since the 1990s, there has been a fundamental change in how innovations have occurred. In-house research and development laboratories are no longer responsible for the bulk of novelties; instead, often buyer-supplier networks or specialised suppliers are. Hence, a novel objective for purchasing arose, namely, (a) to ensure the flow of innovation from suppliers into the buying firm and (b) to establish the conditions and to manage buyer-supplier collaborative innovation processes. 5. Strategic positioning: In a firm where the purchasing volume is of substantial size and where there is a scarcity of suitable suppliers, a further objective for purchasing emerges, namely, to ensure a competitive advantage for the firm by designing and maintaining a performant supply network to which the firm has privileged access, i.e. better access than its competitors in order to achieve competitive advantages. In summary, purchasing refers to the supply of goods and services to the firm, complying to the above named objectives. Purchasing can substantially contribute to a firm’s business success.

4.3 Case Study Purchasing at Volkswagen: Building a Global Leader Volkswagen followed a strategy called “Mach 18”, with the aim to become the largest car manufacturer in the world by 2018. Already in the first half of 2016, this target was reached: no other firm sold as many cars as VW. At the centre of this development stands a revolution in purchasing, which coined what the former speaker of VW, Ferdinand Piëch, called “the third industrial revolution in the automotive industry”. When Piëch took over the chairmanship of the Volkswagen board more than 20 years ago, the company was far away from this goal; rather, he found it in poor condition, producing expensive cars in expensive factories. A traditional manager, at that time, would have tried to cut costs in production. But Piëch was no traditional manager. Rather, he had accumulated a rare experience for board members at that time: purchasing experience. While working for Audi (part of the Volkswagen group), Piëch had also served as chief purchaser. He

50

H. Schiele

explains his key experience, which would later save the Volkswagen group, with the case of tires: The most popular—but by no means only—tire with Audi’s customers was size 195-10. However, this was also true for Daimler. “I only had to calculate: If we just threw out the smaller tires in the Audi 100, we will reach volume leadership against Daimler and BMW, and Audi can largely dictate the driving properties of the tires. This is how the idea was born.” (Piëch 2002, p. 59) By this move, the larger tires became cheaper than the smaller ones, which were then also replaced with the smaller car model Audi 80, again increasing volume. Later, this exercise was extended to other models like the Golf, and the Volkswagen group in the end bought eight times more such tires than its nearest competitor. Two things in this example are remarkable: first is the typical price effect of volume bundling. VW not only used pooling of demand and price comparison and thus saved 15% of the price (in addition to benefits in logistics and production complexity reduction, which should be added in a total cost calculation), but also product improvement was applied, as they could now “dictate the properties of the tyres”. In the end, the idea of volume increase by using the same components with many different models led to the modularisation of the car and the introduction of Volkswagen’s platform strategy. From this background, when taking over the chairmanship with Volkswagen, Piëch’s logical first priority was to improve purchasing and to hire the best purchasing manager available. He chose Ignacio Lopez, who had to be headhunted from the competitor General Motors (which later even went into a lawsuit against its former purchasing manager). Lopez was installed at the board level and fundamentally changed the Volkswagen purchasing organisation. Many elements he discussed later became or are becoming “industry standard”, including the following: • Organisation: Instead of having an independent purchasing department at each factory, a matrix organisation is established, where decisions for the entire company are taken in a joined sourcing committee. Thus, organisational conditions for pooling are established. • Advanced sourcing: Instead of waiting for engineering personnel to design the product, members from the new department of advanced sourcing join the new product development team right from the beginning, conducting value analysis workshops, defining the specification, selecting suppliers but also managing the expensive change orders, which may be forwarded from the technical side but are commercially not very desirable. An important task of advanced sourcing is to avoid over-engineering, which is essentially an organisational issue requiring a dedicated process and structural organisation. • Module sourcing: Instead of buying individual components, pre-assembled modules are bought. Modules are parts consisting of several components

4 Purchasing and Supply Management

51

that typically represent one physical unit and can jointly be installed in the final product and take over a certain technical function. In the automotive example, the cockpit would be a module, consisting of several components like the driving wheel, speedometer etc. Buying entire modules instead of their individual components reduces the depth of production of the OEM (original equipment manufacturer, e.g. the car company) and the number of suppliers, thereby making it possible for purchasing to dedicate sufficient time to each supplier and applying the entire available purchasing tool set. These are only a few of the components of the purchasing revolution; others include global sourcing and its twin risk management, continuous improvement, supply chain optimisation, etc. To conclude the case study on how Volkswagen’s purchasing function led the way to this firm’s ascension, there is one important remaining remark: while most certainly the introduction of a modern purchasing organisation by Ignazio Lopez represented the game-changing first spark, it is worth remembering that Volkswagen also applies the idea of continuous improvement not only to technical components improvement but to the development of the purchasing function as such. Volkswagen has installed the AutoUni, an automotive university in which its executive board functions have their own institutes. The institute of purchasing (alongside its peer institutes representing other corporate functions) administrates a Ph.D. programme, which not only nurtures the next generation of corporate leaders but also constantly pursues the application and development of new purchasing practices.

4.4 Different Purchasing Situations and Different Purchaser Roles (Basic) 4.4.1 Different Purchasing Situations Based on the type of purchased objects and its intended contribution in the buying organisation, several purchasing situations can be distinguished. Each situation requires different skills from the purchaser; hence, different purchasing roles emerge (see Fig. 4.4). A purchasing object (material or service) can be distinguished as either direct and indirect material and project or serial products. Direct materials are those that directly contribute to the final product the buying firm provides. Hence, direct materials are built into and disappear in the final product. Indirect materials are those purchasing objects that do not end up in the

52

H. Schiele

Indirect material

Direct material

Fig. 4.4 Different purchasing situations

spot/project

serial

final product but are needed to support the production of the final good. From a managerial perspective, typically, the logistics of direct and indirect materials differ. For instance, just-in-time models, consignation storage, etc., typically refer to direct material. Hence, purchasers responsible for direct materials may benefit more from knowledge on logistics models than purchasers responsible for indirect materials. Another distinction based on the contribution of the purchased object is whether the product is a one-time buy (spot or project) or repeated buy (serial product). A serial purchasing object is ordered in large quantities of equal units for repeated use in similar products. A spot or a project purchasing object is characterised by a single purchasing situation used only for a specific product. From a managerial perspective, the handling of serial as opposed to project goods differs substantially. For instance, serial products typically are sourced with a frame contract valid for a period of time (e.g., one year), whereas project purchasers typically make single contracts for each object. The relative time spent on strategic sourcing is typically higher with project buying, whereas the operative procurement component is often more important in serial production. Direct materials are often serial materials, but this is not necessarily always the case. For instance, in the case of machine building or shipbuilding, direct materials are sourced in the form of project buying, because each of the final products is unique. Finally, based on the nature of the purchased material or service, a distinction between an existing product and a new product can be made. In the case of an already existing purchasing object, at least one supplier on the market is already offering the required material or service in the required form. In the case of a new product, no adequate supply can be found on the market, and a supplier must be contracted to develop it. From the perspective of the buying firm, an existing product can be a straight re-buy, because it had already been sourced before or it already

4 Purchasing and Supply Management

53

exists on the market but has not been sourced by this firm. With new products, an important distinction can again be made between those new products that only have to be adapted slightly according to the requirements of the buyer and those that have the character of a completely new product development. Again, important differences in management arise based on this distinction, for instance, in controlling. With re-buy objects, the price paid before is known, and the difference towards the new price is called savings in the case that the price is lowered. Savings are more difficult to calculate for newly developed products, since no direct reference price exists.

4.4.2 Role Models in Purchasing Based on the different purchasing situations described above, several roles for purchasers can be distinguished. Often, these roles also differ in their main internal cross-functional partners. Typically, these roles require different skill sets and hence different educational backgrounds or different personnel development efforts: • Operative procurement: is responsible for operational activities, i.e., ordering material and expediting the order. In manufacturing firms, the most common crossfunctional interface partners are in production and (inbound) logistics. • Purchaser for direct materials/serial purchaser: This is the most common role in a manufacturing firm. The serial purchaser is responsible for sourcing direct materials for production, developing a sourcing strategy and selecting and contracting suppliers. Typically, such a purchaser is dedicated to all materials from one category of purchased goods. • Purchaser for indirect materials: This person is responsible for sourcing indirect materials, developing a sourcing strategy and selecting and contracting suppliers. Several variants—each with different skills requirements and different crossfunctional interface partners in the firm—can be distinguished, for instance, service purchaser, investment purchaser and MRO purchaser (maintenance, repair, and overhaul). • Public procurement: Most of the purchasing activities in the public and the private sector overlap. However, concerning the legal framework in particular on contracting issues, substantial differences exist, which result in a specialised job profile for public procurement. • Purchasing engineer: The procurement or purchasing engineer—sometimes also called advanced sourcer—joins a new product development team as a permanent team member and interfaces with other development team members on the one side, and other purchasers responsible for specific materials on the other side. The purchasing engineer’s main interface partner in the firm is the research and development (R&D) function. • Chief purchasing officer (CPO): organises the purchasing department and gives leadership to the purchasers, representing the purchasing function in the board of

54

H. Schiele

directors of the firm. In smaller organisations, the CPO also executes several of the roles subsequently described in “other roles”. • Other roles: Depending on the size of the organisation, a further specialisation of roles is often found, for instance, into purchasing controller (evaluates purchasing and supplier performance, monitors strategy execution), supply risk manager (operates the preventive risk assessment in the supply chain and manages the reactive risk mitigation), purchasing HR agent (recruits purchasers and supports their skills development), systems and strategy (implements and update purchasing IT systems as well as purchasing processes, organises strategy development), supplier development engineer (detached to support suppliers to improve their services), supply chain finance (supports suppliers with favourable financing conditions), sourcing market analyst (conducts market analyses and identifies new suppliers), innovation purchaser (systematically searches for supplier innovations) and others. Depending on the size of the firm, these roles are expected to be executed by one person (in the case of a very small company) or are fulfilled by specially trained individuals. The most common model is a distinction between strategic sourcing and operative buying and a CPO who is given some support from a systems and strategy assistant or group. In high-tech firms, the role of purchasing engineers is gaining prominence.

4.5 The Year Cycle of Purchasing (Basic) 4.5.1 Overview Regular purchasing activities can be depicted in the “purchasing year cycle”. Based on corporate planning that reflects the firm strategy, purchasing plans the supply for materials and services and selects and contracts suppliers (strategic sourcing; steps 1–4 in the category sourcing cycle depicted in Fig. 4.5). Subsequently, these plans are executed (operative procurement step 5), and their performance is evaluated (step 6). While this set of sequential activities in the category sourcing cycle is executed at a category level, another set of activities is executed at the level of the entire purchasing department, hence, the “purchasing department cycle”. The overall success of purchasing activities is monitored through controlling activities (1) that contribute to strategic planning (2). Based on the new plans, process and structural organisation must be adapted, and personnel choices are made (3–6). Then, the cycle, which usually is repeated on an annual base, starts again. The category sourcing cycle is executed by strategic sourcing and operative procurement personnel from a particular category group, whereas the purchasing department cycle is administered by the management of the purchasing department, possibly supported by staff personnel, and refers to all category groups in a feedback process with both the department and external stakeholders.

4 Purchasing and Supply Management

55

The Firm Category Sourcing Cycle

Purchasing Department Cycle

Supply Market

Fig. 4.5 Purchasing year cycle

Fundamental to the understanding of the purchasing year cycle is the understanding of the basic organisational unit of purchasing: the sourcing category (also called commodity group, product family, family of materials or material group). The term sourcing category seems best suited to describe the phenomenon, as it avoids the narrowness of definition implied by the word “material groups”, because they also include services. Likewise, it avoids the misconception implied with the term commodity groups, since sometimes simple market-traded goods (i.e., stocks) like raw oil or wheat are called “commodities”. Sourcing categories are general groups of purchased items, including materials or services of a similar type provided by the same group of suppliers that constitute a single supply market. These groups are not formed based on technical characteristics, product characteristics, tax considerations or other sorting criteria but instead reflect the alternatives in the supply market. Suppliers that could serve as alternatives for each other belong to one sourcing category. If supplies are not administered in such category groups, professional purchasing activities face challenges, e.g., pooling of demand becomes difficult. In a traditional purchasing department, purchasers were often responsible for all type of materials required at a certain location or from a certain group of users. As a consequence, such purchasers cannot develop particular industry expertise because they are responsible for many different materials. Second, the same supplier may have been contacted by different purchasers from different factories but from a single company, and they may have even offered different conditions. Third, the buying firm could not use its full purchasing power if several small quantities were negotiated instead of once a large volume. Because of these and other reasons, category management has been introduced.

56

H. Schiele

4.5.2 Category Management Cycle The following activities are typically executed at category level. (1) Demand identification and planning: The planning of demand, i.e., the aggregation of the expected quantities of required inputs from suppliers based on the forecasted sales of the firm, is needed at the beginning of the purchasing year cycle. For that, an analysis of the past is matched with a projection to the future. One tool used for planning is the spend cube. In a spend cube, it is depicted (a) who in the buying firm bought (b) what product from (c) which supplier. Filling a spend cube requires an IT infrastructure relying on a fully installed ERP system including a materials management module and a finance module as well as the software to extract and process the data. Such data are past oriented. For the future perspective, a sales forecast can be used as a basis to break down the planned sales into required purchase objects. However, not all demand may be planned beforehand; in that case, the process starts with a purchase requisition brought forward by a user throughout the year. (2) Category strategy: For each sourcing category, a strategy is defined, which reflects the targets from the corporate budget planning and defines the reflection of this strategy in the category. Despite all differences, each sourcing category strategy benefits from answering the same basic questions, which Arnold summarised with the following points (Arnold 1997): A category sourcing strategy explains (1) the value creation model (make, buy or co-operate), (2) sourcing object (raw material, assembled component, complete system), (3) supply chain model (stock, demand tailored, just-in-time, etc.), (4) number of suppliers (single source, few/many sources), (5) locational concept (local, cluster, global, currency area based, etc.), (6) pooling concept (how are synergies between the production units leveraged) to which (7) lever selection may be added. Sourcing levers are tactics used to generate purchasing projects and employ the sourcing. Typical sourcing levers include bundling volumes, price evaluation, supply base expansion, product optimisation, buyer-supplier process optimisation or relationship approaches (for details see Sect. 4.8). (3) Supplier identification and selection: Once the need is known and clearly defined, the selection of the best possible supplier is the next step. For that purpose, the purchaser issues a request for quotation or proposal (RFQ), which contains all necessary information for the potential suppliers, such as a technical description of the item, quality levels required, quantities wanted, dates needed, delivery locations, payment terms, etc. Based on the RFQ, suppliers willing to compete for the order submit quotations. Since typically each offer differs along the many parameters defined in the RFQ, these offers must be made comparable. A major and often very time-consuming activity of the purchaser is to homogenise the diverse offers so as to make them comparable. In connection to supplier selection, two tools may be highlighted: preferred supplier lists and global sourcing. A preferred supplier list contains those suppliers that have been pre-selected based on their past performance to serve as

4 Purchasing and Supply Management

57

preferential partners for receiving RFQs. Preferred supplier lists satisfy at least four purposes: (a) they simplify the work by limiting the scope of search and relying on suppliers that are already familiar with the requirements of the buying firm; (b) they ensure that purchasing volume is channelled to the pre-selected business partners with whom volume discounts can be achieved; (c) they avoid the uncontrolled inflation of suppliers serving a firm, which cannot be properly managed and risk controlled if their numbers grow excessively and (d) they create an incentive for suppliers to perform well in order to be listed on or remain in the preferred supplier list. In case no or too few potential suppliers are available, it can make sense to run a global sourcing exercise. Global sourcing refers to the identification and possible contracting of suppliers located in other countries than the buying company. The association herewith is typically the following: by identifying previously unknown suppliers in low-wage countries, the buying company may (a) profit from their factor cost advantages and (b) increase competition among their existing supply base. The challenge with global sourcing, though, is that in addition to the price, further costs may arise. Therefore, it might prove beneficial for firms to adopt a total cost perspective exceeding a price perspective, only. (4) Negotiation and contracting: Once a shortlist of potential suppliers is defined, an often intensive negotiation process starts on which end one or more contracts are signed (depending on the decision to apply a single source tactic or use multiple sources). Negotiation is a formal process of communication in which the different parties seek to reach a mutual agreement about an issue, in case of a supplier negotiation about the terms and conditions of a purchasing contract. Negotiating differs from bargaining since it is a multi-dimensional exercise that is not solely directed at price comparison. It is commonly agreed that the negotiation success depends strongly on professional preparation, which requires a good understanding of the position and expectation of the supplier as well as of the buyer’s own targets and limitations. A buyer may need to define its LAA (least acceptable agreement), its MDO (most desired outcome) and its BATNA (best alternative to no agreement). In case of perfect information in a fully transparent market, negotiations would not be necessary. Markets coming close to this status typically develop stock exchanges in which homogenous goods are exchanged and form a market price, which is equal to all. Once an agreement is reached, contracts are signed, i.e. legally binding agreements resulting from an offer and its acceptance that specify the terms and conditions of the transaction. The Incoterms (international commerce terms, updated since 1936) can be used to clarify conditions, for instance, by specifying whether the cost and freight are covered (CFR), or if conditions apply ex works (EXW), and the buyer collects the purchased goods at the supplier’s location. There are many types of contracts, such as fixed-price or cost based contracts, short or long term contracts and partnering agreements. Annual contracts are often used, which are frame agreements that cover purchases for the next year.

58

H. Schiele

(5) Executing: This is also known as “purchase-to-pay process”. Once a supplier is contracted, purchasing orders can be placed. Depending on the purchase situation, the contract may in fact contain the purchase order, whereas in annual contracts, for instance, call-off agreements will be used against which material requisitions are placed. In the case of serial production, this is typically done automatically through connected IT systems, once the sales confirmation for the final product has been recorded or, alternatively, once a pre-defined minimum inventory level is undercut. Once an order is placed, confirming, follow-up and expediting activities, in case a shipment is threatening to be overdue, are executed in order to ensure that the items are delivered according to the specifications and on time. Here, a main activity of operative procurement is concentrated. Finally, the invoicing and payment process concluding a transaction is part of the execution. The productivity of the operative procurement process has been substantially improved in the last years through automation. (6) Supplier evaluation: At the end of the process, supplier evaluation may take place, i.e., a systematic assessment of a supplier’s performance after delivering one or more purchased items. As opposed to the previous steps—in particular, supplier selection, contracting and execution—supplier evaluation is not technically necessary to apply or legally binding. But it is an integral step in the purchasing process of a mature buying organisation. Two general types of supplier evaluation can be distinguished: quantitative and qualitative supplier evaluation. For quantitative supplier evaluations, data are retrieved from the IT system and cover metrics such as delivery reliability, quality complaints, returns rate, etc. Quantitative supplier evaluation data are easier to generate since they come directly from the ERP system. This data mainly documents the supplier’s performance but does not explicitly note reasons for potential problems. To document reasons for potential problems, a qualitative supplier evaluation is beneficial. Here, buyer employees having contact with the supplier are asked to fill in and discuss among each other a set of questions concerning the perceived functioning of the supplier (e.g., performance, costs, service and soft facts, systems and strategy). In an industrial setting, qualitative supplier evaluation questionnaires are usually filled in by strategic sourcing, operative procurement/logistics/production, quality and R&D personnel. These pieces of information may be used to include or to remove a supplier from the preferred supplier list or to start supplier development programmes in order to overcome identified weaknesses. With supplier development, an active and a passive form can be distinguished. In active supplier development, the customer devotes their own resources to support the supplier (e.g., sending quality personnel, inviting supplier personnel to trainings, etc.). In passive supplier development, the supplier is asked to improve by itself (e.g., committing to improve the weak points detected in supplier evaluation and checking the progress after a determined period). Qualitative supplier evaluation typically takes place once each year and thus concludes the category management year cycle.

4 Purchasing and Supply Management

59

4.5.3 Purchasing Department Cycle Whereas the above described activities of the “category sourcing cycle” (see Fig. 4.5) are executed at the category level, the following activities of the “purchasing department cycle” are administrated at the departmental level by the CPO and staff employees. (1) Controlling: Purchasing controlling fulfils three important functions by (a) calculating and monitoring savings, (b) administering project progress and (c) preparing performance and improvement reports. In the annual budgeting process of the firm, the size and development of the purchasing volume plays an important role in achieving the profit targets. This overall budget target must be broken down to each category group’s contribution. Out of this plan, target costs can be derived (see Fig. 4.6). Often, a difference between past costs and future allowed costs may arise; hence, purchasing must generate savings. Savings can be calculated against past costs (e.g., against first offers from suppliers, against price indices) or against target costs that fit into the overall corporation planning process. Controlling helps to define targets and monitors the achievement throughout the year. Project monitoring is closely linked to this activity. In order to achieve their targets, the responsible purchasers will define and subsequently execute sourcing projects that enable the achievement of the targets. These projects (e.g., enabling a new supplier, realising process cost improvements with a supplier, reducing costs by improving the sourced product) need to be monitored throughout the year. For this, a tool called degree of implementation (DI) has been suggested. DI is a project monitoring metric that splits project progress into five clearly defined Increase sales

Reduce costs

Target profit

! Savings of X € needed Turnover old

Costs old

Change Other in cost product reducti on

Basis last year

Fig. 4.6 Annual planning

Purcha- Costs Volume sing with change cost conreduc- stant tion quantity

Target costs

Target Promised turnprofit over

Target next year

60

H. Schiele

Level 1: Firm strategy

Firm A

Level 2: Functional strategies e.g. Control unit

Level 4: Sourcing levers (tactical) Level 5: Supplier strategies

Purchasing

Other functional strategies

Other category strategies

e.g. Display

Sourcing category

Level 3: Category strategies

e.g. Engineering

e.g. Process improvement

e.g. Price evaluation

Other sourcing levers

Supplier A

Supplier B

Other supplier strategies

Implementation Category performance Purchasing performance Business performance

Fig. 4.7 Hierarchy of strategies in purchasing (Hesping and Schiele 2015)

execution levels that can be reported. The five levels are (DI1) idea defined, (DI2) savings estimated, (DI3) initiatives defined for implementation, (DI4) savings contracted and (DI5) savings financially effective in the balance sheet. Next to costs, controlling monitors the achievement of the other purchasing targets, such as its contribution to innovation or to the strategic goals of the company. (2) Strategising: A purchasing strategy must be aligned to and often be guiding for the overall strategy of the corporation. There is one important thing to remark, however: purchasing may better be seen as relying on a hierarchy of strategies (see Fig. 4.7). The reason for this is lying in the organisational principle of purchasing departments, which typically are structured along categories of goods. Each category may be in a different purchasing situation, for instance, direct or indirect material requiring different approaches (see Fig. 4.4). Hence, each category needs its particular strategy. Jointly, these individual category strategies contribute to the achievement of the firm’s overall targets. (3) Process organisation: Organising includes the establishment of the structures of a firm and the design and implementation of a supportive process organisation. Here, typically a strategic sourcing process is specified next to an operational purchase-to-pay process, the former covering steps 1–4 in the category sourcing cycle and the latter step 5. Depending on the industry, the level of maturity and the size of the firm, purchasing activities may recur on more specific processes, such as an early supplier inclusion process in new product development projects, a supply risk management process, a supplier development process and others. (4) IT process support: An important tool here is e-procurement. E-procurement refers to the digitalisation of procurement processes, electronically linking buyer

4 Purchasing and Supply Management

61

and supplier. A common form uses e-catalogues, which contain negotiated items and allow the end users to order materials. EDI (electronic data interchange) is another popular technique that allows for a direct and digital data exchange between buyer and supplier. It is widely expected that as the consequence of the fourth industrial revolution (industry 4.0) with cyber-physical systems and autonomous machine-to-machine communication, the automation of operational procurement activities (but possibly also parts of the sourcing process such as digital negotiations) will progress. (5) Structural organisation: Following the classical dictum of “structure following strategy”, the structural and process organisation of the purchasing function must be reviewed annually. An enduring question of organisational structures is balancing the level of centralisation with decentral authority. This issue is of particular relevance in purchasing, as from a pooling perspective, a group of companies would preferably have one central purchasing department only. Based on the amount of involved persons or locations in an organisation and the extent of the formalisation of the product, purchasing literature and practice has, however, come with three suggestions on how to cope with this tension in an intermediate way, i.e., neither fully centralising nor fully decentralising purchasing: implementing lead buyers, purchasing councils or shared services. Lead buyers are individual purchasers responsible for buying a particular category for the entire group. This model is recommended if a particular location of a multi business unit corporation has a high percentage of the total purchasing volume. In purchasing councils, delegates from each business unit jointly negotiate contracts valid for all units. This model is often applied if 4–8 business units with similarly sized demand are present. Shared services, finally, refer to an organisational model that aims at providing corporation-wide support in a particular category by pooling services together and installing a special organisational unit. Shared services are considered to be most effective if there are many users with a similar, fully specified demand. (6) Staffing: Finally, once targets have been defined and processes and structures have been adapted to make the attainment of these targets possible, adequate personnel must be employed, respectively trained, to fulfil the required tasks. Here, the role models explained in Sect. 4.4 can be used. Supported by a suitable organisational structure and processes that enable the department to reliably produce similar results, the cycle can start again.

4.6 Theories Supporting Purchasing Decisions (Advanced) Purchasing is a field of business administration. Business administration science, in turn, has traditionally been application oriented; that is, it attempts to provide industrial practice with well-founded but implementable decision support. At the same time, a shared theoretical fundament for purchasing science as a means for providing

62

H. Schiele

systematic decision support is just about to be developed. A literature review reveals that about two-thirds of the covered academic papers on supply management does not rely on any particular “grand theory” (Chicksand et al. 2012). Those using theory to base their assumptions typically rely on transaction cost theory, principal agent theory and resource based/network theory models. This analysis reveals another interesting finding: papers that relied on a particular theory were cited significantly more often than those not embedded in any particular school of thought. This indicates that the development of a field may speed up by relying on a common set of theories that are gradually better understood. The rationale behind this observation may be that theory-based research avoids redundancies, because it is easier to understand what has been previously studied. Therefore, some of the most applicable theories for purchasing decision support are briefly sketched below. The measure for “applicability” and theory selection here is not only its past application in literature, but first of all the potential to solve decision problems in the purchasing year cycle. The main decisions purchasers responsible for a category group must make in the year cycle can be bundled in four main categories: (1) the make-or-buy decision needed to determine if a purchasing act will be necessary; (2) the decision on a suitable sourcing strategy for the entire category group, for instance, determining the number of suppliers; (3) the decision on a particular strategy towards the individual suppliers chosen as preferred within a category group, defining how the relationship should be designed, for instance, in a more collaborative way or on arm’s length terms and, finally, (4) support for the final decision to negotiate and sign a contract in the interaction with each supplier. The ideal theory would offer guidance for all decision points. However, there is no such thing as a universal purchasing theory available, so far, but only partially supportive theories: • Transaction cost economics: Until now, the most popular theory backing purchasing research relies on the work of Coase and the subsequent development of the theory by Williamson, for which both were awarded a Nobel prize. In essence, the theory starts with the position that each economic transaction (e.g., a buyer-supplier exchange) incurs particular costs called the transaction costs (Williamson 2008). Depending on their extent, firms should either produce their products in-house, outsource the production or possibly find a position in between the two extrema. Transaction cost theory is based on the assumptions of bounded rationality and opportunism. From this theory, an apparent contribution to the make-or-buy decision can be drawn (decision point 1). The theory may also contribute to decision point 4 (contracting) by elaborating on the function of contracts and contractual safeguards contributing to reduce transaction costs. • Resource dependency theory: This theory is based on a book called “The external control of organizations” published by Pfeffer and Salancik (Hillman et al. 2009; Pfeffer and Salancik 1978). The title of this book already gives an indication of the close link with purchasing, the latter being the function responsible for the management of the external supply links of an organisation (even though the book lacks a clear supply management focus). Resource dependency theory, in

4 Purchasing and Supply Management

63

essence, sees the firm as an open system and posits that firms should reduce their dependence on external resources controlled by other parties. The ability to acquire and maintain external resources is seen as essential to a firm’s survival. Again, assuming bounded rationality, this theory focusses on contributing to decision points 2 and 3, formulating category group strategies as well as individual strategies towards suppliers. • Relational view of the firm: The only “grand theory” that may claim to have its origin in purchasing research is Dyer and Singh’s “relational view of the firm”, which can be seen as a special form of resource-based argumentation, focussing not on internal but on external resources (Dyer 1996; Dyer and Singh 1998). The authors studied the differences between Japanese and American buyer-supplier relations in the automotive industry. According to this theory, firms may obtain relational rents by collaborating in the supply chain. Relational rents are returns on collaboration between independent partners who jointly are able to produce better results than alone. Relational rents are determined by the availability of relation-specific assets, knowledge sharing routines, complementary resources and effective governance structures. The importance of this theory is further underpinned as it contradicts the verdict of the classical market-based view in strategic management, which assumes perfect factor mobility, i.e., equal access to suppliers by all market participants and thus no strategic contribution by purchasing (Mol 2003). Apart from the general positioning of purchasing as a strategic task with the target of providing competitive advantage to the firm by generating relational rents together with suppliers, the relational view concretely contributes to the formulation of category strategies and supplier strategies (decision points 2 and 3). • Principal agent theory: Agency theory shares similarities with transaction cost theory but focusses on contracts and relational agreements between a principal (e.g., the buying company) and an agent (e.g., the supplier) and has found substantial application in supply chain management (Fayezi et al. 2012; Steinle et al. 2014). Especially at the beginning of a new relationship, potential benefits of trust are challenged by the potential risks of opportunistic behaviour fostered through asymmetric information. Thus, the chance of the supplier to act opportunistically is considered as a relevant factor affecting the buyer–supplier relationship. Agency theory suggests various possibilities (e.g., signalling, screening, monitoring) to effectively limit this opportunistic behaviour. It becomes clear that here again a theory is presented that can benefit purchasing, in particular, in deciding on the design of a buyer-supplier relation and contracting (decision points 3 and 4). • Social capital/exchange theory: Buyer-supplier interaction does have a strong relational component. Social capital and social exchange theory focus on this relationship. Social exchange theory assumes that market exchange may involve both economic and social outcomes; it develops a new perspective on opportunism, and it rejects a universal homo oeconomicus, strictly maximizing economic profits (Lambe et al. 2001). Thus, the assumption that economic agents behave opportunistically whenever the possibility arises is refuted. Instead, social exchange theory analyses under which conditions a relationship is initiated (if the partner is considered to be sufficiently attractive) and when it is continued (if the

64

H. Schiele

relationship satisfies the expectations, and no better alternative is available). Social capital theory, then, specifies a relation by distinguishing between structural, relational and cognitive forms of social capital. These theories can be helpful to describe buyer-supplier relations (decision point 3) but also to develop category strategies and design supplier portfolios. • Game theory was originally created by von Neumann, when he developed applicable maths to calculate the best solution for strategy games and then made the link to business application. Essentially, game theory attempts to determine all possible scenarios in a game, after which the best strategy should be found for each player, whose actions in turn influence the other player(s). The underlying assumption here is the rationally acting homo oeconomicus, who assesses his options and makes the choice with the highest predicted outcome. Concerning the application to purchasing decision support, buyer-supplier negotiations can be interpreted as games (Nagarajan and Soši´c 2008). Likewise, game theory may support sourcing strategists when analysing the supply market. Closely linked, “mechanism design” is a field in which actors try to design a (for instance, negotiation) environment to achieve the desired outcome of a game (decision point 4). It is important to note that the above selected theories cannot all be applied jointly, because some rely on contrasting assumptions, such as bounded rationality as opposed to the rational homo oeconomicus. In conclusion, it can be stated that there are theories available that help to underpin purchasing decision-making and can serve as guidelines not only for practical decisions but also for future research. However, a single “general purchasing theory” is not available.

4.7 The Kraljiˆc Matrix and the Development of Sourcing Strategies (Advanced) The Kraljiˆc Matrix is probably the most famous and widely applied tool in purchasing and is named after its inventor. The 2 × 2 ‘Kraljiˆc matrix’ (Kraljic 1983) distinguishes a noncritical, leverage, bottleneck and strategic portfolio quadrant along two dimensions: (1) ‘strategic importance’ and (2) ‘supply risk’. For each portfolio quadrant, literature offers corresponding, generic strategic and tactical recommendations (see Fig. 4.8). Peter Kraljiˆc was a young consultant on a major project of one of the German chemical giants when he was asked to work with the company’s purchasing department. At that time, the portfolio approach in strategic management was popular. This approach argued that firms should only be active in businesses that showed a growing market and in which they had a strong position, preferably among the top three. Inspired by this idea, Kraljiˆc created an approach to purchasing strategy, which he called the “purchasing portfolio analysis”. The firm was so thrilled by the model that it took four years till they allowed a first publication (Kraljic 1977)

4 Purchasing and Supply Management

65

high

Strategic importance (volume purchased, % purchasing volume, impact on product)

Leverage categories Exploitation of full purchasing power

Strategic categories Development of long term relationships

Noncritical categories Product standardization and efficient processing

Bottleneck categories Volume insurance (at cost premium if necessary)

low low

Supply risk (availability, amount suppliers, competitive demand, substitutions)

high

Fig. 4.8 The Kraljiˆc matrix (Based on Kraljic 1983)

The following steps are suggested to implement his approach: 1. Classification of category group: First, the strategic situation for each category group is analysed. Four general situations are distinguished, depending on the products’ impact on the profit of the buying firm and on the supply risk. 2. Systematic analysis of the supply market: This analysis is used to determine the bargaining power of a buyer and its suppliers. Note that while the first step was conducted at the category group level, now the supplier level is addressed. Based on criteria like suppliers’ capacity utilisation and the uniqueness of its product, the strength of the supplier is analysed. Annual volume purchased and expected growth as well as demand stability are, then, used as criteria to analyse the attractiveness of the buyer and hence the strength of its bargaining power vis-à-vis suppliers. 3. Strategic positioning and strategy development: Based on the information from the previous steps, for each category a strategy is developed. For that, Kraljiˆc recommends standard strategies. Typically, for strategic commodities long-term supplier relationships that enable differentiation towards the competitors are recommended. For bottlenecks, the focus should lay on ensuring supply and possibly replacing the product. With leverage goods, short-term supply and aggressive price negotiations are recommended and with non-critical items, typically efficiency in operative procurement would be the solution. 4. Action plans: Finally, projects with measurable targets, milestones and assigned responsibilities have to be developed for each category.

66

H. Schiele

With Kraljiˆc’s work, a first comprehensive strategic sourcing approach has been presented; this greatly progressed purchasing practice and moved it from an operative case-by-case procurement approach to a strategically planned and coherent sourcing approach. His category descriptions—leverage, bottleneck, strategic and routine—have entered into the standard vocabulary of purchasers. However, the matrix was also criticised. For instance, the model has been condemned because of its lack of analytical rigor: it ignores important influencing factors in determining a buyer’s profit impact and supply risk (Cox 2004). This, however, does not necessarily falsify the general idea, but offers a path to refine the assessment. Another difficulty arises from the challenge to apply the model. Sometimes the level of analysis—category or supplier—is unclear. Kraljiˆc uses two steps, the positioning of the category and then the analysis of buying company and suppliers’ strength. Often, these two steps are merged into just one matrix, which may be somewhat simplistic, as it neglects the role of individual buyer-supplier relations. Another criticism refers to his recommendation of standard strategies, which appear to be exclusive for each situation. Empirical work, however, has shown that sourcing tactics are used more in an additive rather than an exclusive way (Hesping and Schiele 2016a). Finally, Kraljiˆc has been criticised for not sufficiently detailing what action plans should finally look like. Here, a potential remedy is the sourcing “lever” approach.

4.8 Lever Analysis and Cost Savings (Advanced) Another important tool in strategic sourcing that fills the gap between category strategies and actionable plans is the lever analysis (see Fig. 4.7). Sourcing levers describe sets of tactics used to operationalise sourcing strategies as a combination of coherent activities in a sourcing category. The levers are the “tricks” used by purchasers to achieve cost savings. A lever analysis exercise ideally is executed by a category team, organised by the responsible purchaser, but includes the main cross-functional partners, such as quality assurance, logistics/production and engineering/R&D. Jointly, they systematically analyse the options best suited to implement a sourcing strategy and to generate cost savings by iteratively analysing each tactical lever; eventually, they merge a coherent set of levers into an action plan. Following the lever analysis approach, each individual sourcing category is checked to determine which levers could best be applied. In the literature, diverse lever models can be found, ranging from five to 114 levers. Most of them, however, can be grouped into any of the following seven types (Hesping and Schiele 2016b): 1. Volume bundling refers to the consolidation of demand and increasing the purchase volume for quotation. Instead of buying similar items separately from each

4 Purchasing and Supply Management

2.

3.

4.

5.

6.

7.

67

other at each location, a firm maximises its purchasing power by pooling demand and buying jointly. Price evaluation refers to forming price targets and analysing suppliers’ bids and cost structures. Application of novel forms of price negotiation such as auctions also fall under this lever. Extension of supply base refers to increasing the number of sources and bidders per request for quotation to raise bargaining power; this often involves global sourcing projects but also supplier development to nurture new competitors. Product optimization refers to modifications to the design, functions and materials of the purchased items. For instance, a design that requires less material generates savings and does not harm the supplier. Process optimization refers to efficient and effective processes related to the buyer-seller interfaces. Often, automation in operative procurement and industry 4.0 applications, but also in logistics can be applied here. Optimization of supply relationship refers to establishing and maintaining a longterm, mutually beneficial, privileged relationship between buyer and supplier. Partnering contracts and supplier alliances can be a case at hand. Category-spanning optimization refers to balancing trade-offs between multiple sourcing categories and enforcing mutual approaches from otherwise distinct sourcing teams.

Levers 1–3 are also called “commercial levers”, because they can be applied by purchasers alone and have limited support from other functions, whereas levers 4–7 are known as “cross-functional levers”, because their applications require intensive collaboration with other functions. For instance, product optimisation links to engineering, whereas process optimisation often links to logistics. The commercial levers try to exploit existing benefits, while the cross-functional levers try to explore new benefits. Empirically analysing the application and effect of sourcing levers, it has been found that several levers are typically employed in each category. It has also been found, however, that not all levers can be applied at once; hence, there is a need to first draft a category strategy. For instance, there appears to be a relevant trade-off between global sourcing and product optimisation (Schiele et al. 2011). With this set of empirically evidenced sourcing tactics, purchasers have a set of tactics at their disposition that enable them to systematically leverage the sourcing potential of a firm and to link their category strategy to actionable purchasing projects. Lever analysis has been criticised for its lack of academic rigor, as its origins are, similar to the case of the Kraljiˆc matrix, in practice. While this may be true for some models, it does not falsify the entire approach. Another criticism refers to the general attitude of addressing individual categories instead of, for instance, entire products or processes. This is a fundamental concern challenging the implementation of category management in purchasing. To mitigate this criticism, the seventh lever (“categoryspanning optimization”) has been introduced, which explicitly asks to consider the effects that the optimisation of one category may have on others (e.g., cheap copy paper may require more printing ink). Lever analysis may be best suited to satisfy the

68

H. Schiele

cost savings target of purchasing, but it may contribute less to the other targets such as innovation and strategic positioning. Finally, the efficiency of sourcing levers may be limited by the level of purchasing maturity an organisation has (Schiele 2007). The sourcing levers are tactical in nature. For a strategic (re-)positioning of a firm, other tools, like the preferred customer matrix are more supportive.

4.9 Achieving Preferred Customer Status and Supplier Satisfaction (State-of-the-Art) 4.9.1 Preferred Customer Policy as Means to Achieve Competitive Advantage The last two decades revealed two fundamental changes occurring in the supply chain: first, a concentration on core competencies and the outsourcing of the remaining functions steadily reduced the depth of production of industrial firms. This trend increased the importance of purchasing in general (see Fig. 4.1). Second, and in parallel, in purchasing the trend prevailed to reduce the number of suppliers and concentrate on a few close buyer-supplier relations. Thus, in many industries, the number of available suppliers sunk, often causing oligopolistic situations, while their importance increased. This trend challenges purchasing to react with novel approaches. Firms may need to become the preferred customers of the few remaining world class suppliers. For that purpose, they may benefit from taking a different perspective on buyer-supplier relations and work hard to have satisfied suppliers. A firm has preferred customer status with a supplier if the supplier offers the buyer preferential resource allocation, i.e., better access to its valuable products or services than it offers to other customers. This can be accomplished in several ways. A supplier may dedicate its best personnel to joint new product development projects, customise its products according to the customer’s wishes, offer privileged treatment if bottlenecks in production occur and offer innovations first or even enter into an exclusive agreement (Steinle and Schiele 2008). The core assumption here is that not all customers are treated equally, because suppliers have to make a choice in view of resource scarcity. Most of the targets of purchasing benefit from a buying firm enjoying preferred customer status: in particular, in cases of supply shortage, safe supply is provided to the preferred customers, while other customers may suffer from a supply interruption. Preferred customer status reduces supply risk. Further, research has found indications that suppliers offer beneficial pricing to their preferred customers, as they appreciate their consistent business. Likewise, success in buyer-supplier collaboration for innovation is strongly influenced by the customer’s status with the supplier. Finally, with the preferred customer approach, purchasing has a chance to satisfy the novel target of contributing to a firm’s competitive advantage. Having exclusive

4 Purchasing and Supply Management

69

access to a supplier with valuable capabilities creates a strategic advantage for this firm. The question arises: How can a buying firm achieve preferred customer status with its suppliers? Here, social exchange theory can be used to explain the phenomenon and to offer suggestions on how to improve the standing with suppliers.

4.9.2 Social Exchange Theory: Supplier Satisfaction as Antecedent to Preferred Customer Status Social exchange theory analyses the establishment and development of social relations, such as buyer-supplier relations. First, the potential partners need to be sufficiently attractive to each other to start a relationship. Then, they have certain expectations towards this relation, against which they assess the relation after having gained sufficient experience. Interestingly, then, social exchange theory introduces the “comparison level of alternatives”. According to this theory, in a third step, the partners compare their satisfaction with this particular relationship with other potential alternatives. Only then actors decide to continue the relation or not. They may thus categorise a business relation into standard, preferred or, in the worst case, exit. In buyer-supplier relations, thus, buying firms (1) need to be sufficiently attractive for potential suppliers to get a quotation from them. In case a business relation is established, then, (2) the supplier needs to be satisfied with the relation. Finally, (3) the supplier needs to be more satisfied with this customer than with its alternatives so that the buying firm finally becomes a customer of choice and hence get privileged treatment. From a buying firm’s perspective, two activities need to be performed: (1) Obtain a notion of one’s own status with the suppliers and adjust strategies accordingly. For instance, it may not be a good investment to run supplier development activities with suppliers who see the customer as an “exit” customer. (2) Understand if suppliers are satisfied and improve their satisfaction. In order to assess a buyer’s strategic situation with the suppliers, the “preferred customer matrix” has been developed (see Fig. 4.9). One of its axes depicts the status a buyer has with a supplier, and the other axis shows the competitiveness of the supplier. A supplier’s competitiveness typically is assessed in a very companyspecific way, depending on the strategic direction of the firm. Suggested criteria to assess the other axis, a firm’s status with its supplier include: • technical match (strategic importance of the customer firm for the supplier due to a congruence of technological roadmaps), • commercial importance (significance of purchasing volume in supplier’s overall business), • cultural fit (existence of similar cultural values in buyer and supplier firms), • past preferential treatment (evidence of preference in supplier’s past behaviour), and

Squire “

• Supplier development strategy—Work with supplier to increase its competitiveness

Quacksalver “



• Replacement strategy—Seek out new suppliers that offer some advantage

King “

• Collaboration strategy— Work with supplier to achieve competitive advantage





Standard customer Preferred customer

H. Schiele



Buyer´s status with supplier

70

Black Knight “

• Bonding strategy—Pursue concerted program to earn preferred customer status with supplier

low

high Competitiveness of supplier

Fig. 4.9 The preferred customer matrix (Schiele 2012)

• key account status (awarding of key account status with the supplier’s sales, R&D, quality, and production departments). After mapping all A and B suppliers (classically understood as those who jointly are responsible for 80% of the total purchasing volume), and also mapping selected smaller suppliers that are considered to be important for diverse reasons, the overall situation of the firm is first analysed. For instance, it is not very beneficial if the bulk of suppliers have to be ranked as “black knights”, i.e. the most competitive suppliers awarding a firm’s competitors preferred customer status. In this situation, the firm is strategically vulnerable, because the best suppliers prefer to work with its competitors. In such a situation, a corporate strategy of being a technology leader may not be feasible. Here, we see how the supply situation can enable or prevent the strategy of the entire firm. Second, after the overall assessment, a particular strategy is developed for each individually mapped supplier, which may include the attempt to change its position in the matrix or to possibly discontinue the relationship in case of a “quacksalver”, which is neither particularly competitive nor shows appreciation for the buying firm. Finally, what can a firm do once it has detected a problem in its supplier portfolio or, more generally, when it wants to improve its overall standing with its suppliers? For that purpose, a supplier satisfaction analysis can provide valuable insights. Current research has revealed several main factors that explain supplier satisfaction. These can be operationalised in order to understand if suppliers are satisfied with a particular customer:

4 Purchasing and Supply Management

71

• Growth opportunity: For sales personnel, it is more interesting to collaborate with a customer whose turnover is growing, so that in a subsequent period more products can be sold. • Profitability: Obviously, a supplier is more satisfied if a customer is paying fair. • Relational behaviour: The behaviour of a customer’s personnel in terms of reliability, support offered or openness for supplier involvement influences the supplier’s satisfaction with this relationship. • Operational excellence: Operational excellence like simple processes, prompt responses and accurate forecasts also positively influence the relationship. In order to obtain valid results, supplier satisfaction is often assessed not at an individual level but at a corporate level and with a neutral intermediary like a university executing the survey. With this result, the firm can then systematically improve its position with suppliers by working on its identified weaknesses. For instance, if suppliers are unsatisfied with the corporate planning, the buying firm may start a project to improve its planning processes. If, on the other hand, suppliers do not see a growth perspective, the buying firm may start an upstream marketing campaign that better explains its future prospects to suppliers, etc. In summary, the preferred customer perspective reaches novel conclusions. For instance, the most highly regarded global supplier may not be the best supplier for a particular buyer that may not have any chance to achieve preferred customer status. Essentially, a shift in perspective is implied. Reflecting the improvement of the position of oligopolistic or generally strong suppliers, purchasing may have to work hard to become an attractive partner for suppliers rather than waiting for suppliers to queue to offer their services. Firms that first apply this approach in a particular supply market becoming more attractive as customers than their competitors since they may act as a “game changer” and strategically outperform their competitors. Practically, the first large firms have established “supplier club programmes”, in which a selected number of preferred suppliers can obtain a special set of privileges with the intention to ensure them awarding the buying firm with preferred customer status. Knowing and improving its strategic position with suppliers is also a first step for purchasing to genuinely contribute to corporate strategic planning. While the suppliers’ importance is quantitatively apparent, so far few instruments have been available to measure and operationalise purchasing’s strategic contribution. The research stream on preferred customer status may thus pave the road to a next step in purchasing maturity, enabling firms to successfully manage the new situation of a network economy with few and strong suppliers and gain competitive advantage from its positioning in the supply chain.

72

H. Schiele

4.10 Further Reading A conceptual framework for the field of Purchasing and Supply Management is provided by Kaufmann (2002). Kirkman (1887) is an excellent source on the history of supply management for the interstate US railways, while Redtmann (1910) is an early reference on the structuring of a purchasing department. Van Weele (2005) provides a good overview of purchasing and supply chain management at both a strategic and an operational level. Details on the importance of purchasing for Volkswagen can be found in the autobiography of Piech (2002), see also Versteeg (1999). A thorough discussion on a categorization of purchasing situations and the roles taken by purchasers is provided by Hesping and Schiele (2015), Schumacher et al. (2008) and Arnold (1997). Theoretical foundations and in particular the discussion of the lever models can be found in Hesping and Schiele (2016b), based on the use of the Kraljic matrix which was initially published by Kraljic (1977, 1983), see also Caniels and Gelderman (2005), Cox (2014), Schiele et al. (2011) and Schuh et al. (2009). An analysis of the importance of a customer preferential position at the supplier is discussed in depth by Schiele et al. (2012), Schiele (2012) and Vos et al. (2016).

References Arnold U (1997) Beschaffungsmanagement, 2., überarbeitete und erweiterte Auflage. SchäfferPoeschel, Stuttgart Baily P, Farmer D, Crocker B, Jessop D, Jones D (2008) Procurement principles and management, 10th edn. Pearson Education, Harlow etc. Caniels MCJ, Gelderman CJ (2005) Purchasing strategies in the Kraljic matrix—a power and dependence perspective. J Purch Supply Manag 11(2–3):141–155 Chicksand D, Watson G, Walker H, Radnor Z, Johnston R (2012) Theoretical perspectives in purchasing and supply chain management: an analysis of the literature. J Purch Supply Manag 17(4):454–472 Cox A (2004) Business relationship alignment: on the commensurability of value capture and mutuality in buyer and supplier exchange. Supply Chain Manag Int J 9(5):410–420 Cox A (2014) Sourcing portfolio analysis: power positioning tools for category management & strategic sourcing. Earlsgate Press, London Dyer JH (1996) Specialized supplier networks as a source of competitive advantage: evidence from the auto industry. Strateg Manag J 17:271–291 Dyer JH, Singh H (1998) The relational view: cooperative strategy and sources of interorganizational competitive advantage. Acad Manag Rev 23(4):660–679 Fayezi S, O’Loughlin A, Zutshi A (2012) Agency theory and supply chain management: a structured literature review. Supply Chain Manag Int J 17(5):556–570 Hesping FH, Schiele H (2015) Purchasing strategy development: a multi-level review. J Purch Supply Manag 21(2):138–150 Hesping FH, Schiele H (2016a) Matching tactical sourcing levers with the Kraljiˇc matrix: empirical evidence on purchasing portfolios. Int J Prod Econ 177:101–117. https://doi.org/10.1016/j.ijpe. 2016.04.011 Hesping FH, Schiele H (2016b) Sourcing tactics to achieve cost savings: developing a formative method of measurement. Int J Procure Manag 9(4):473–504

4 Purchasing and Supply Management

73

Hillman AJ, Withers MC, Collins BJ (2009) Resource dependence theory: a review. J Manag 35(6):1404–1427 Johnson PFL, Michiel R, Flynn A (2011) Purchasing and supply management, 14th edn. McGrawHill College, New York Kaufmann L (2002) Purchasing and supply management—a conceptual framework. In: Hahn D, Kaufmann L (eds), 2., überarbeitete und erweiterte Auflage ed. Gabler, Wiesbaden. pp 3–33 Kirkman MM (1887) The handling of railway supplies: their purchase and disposition. CN Trivess, printer Kraljic P (1977) Neue Wege im Beschaffungsmarketing. Manag Mag 11(11):72–80 Kraljic P (1983) Purchasing must become supply management. Harvard Bus Rev 61(5):109–117 Lambe CJ, Wittmann CM, Spekman RE (2001) Social exchange theory and research on businessto-business relational exchange. J Bus–Bus Mark 8(3):1–36 Mol MJ (2003) Purchasing’s strategic relevance. J Purch Supply Manag 9:43–50 Nagarajan M, Soši´c G (2008) Game-theoretic analysis of cooperation among supply chain agents: review and extensions. Eur J Oper Res 187(3):719–745. https://doi.org/10.1016/j.ejor.2006.05.0 45 Pfeffer J, Salancik GR (1978) The external control of organizations. Harper and Row, New York Piech F (2002) Auto. Biographie, 3rd ed. Hamburg Redtmann C (1910) Das moderne Einkaufsbureau im Fabrikbetriebe. Zeitschrift für Handelswissenschaft und Handelspraxis 2(1910):55–62 Schiele H (2007) Supply-management maturity, cost savings and purchasing absorptive capacity: testing the procurement–performance link. J Purch Supply Manag 13(4):274–293 Schiele H (2012) Accessing supplier innovation by being their preferred customer. Res Technol Manag 55(1):44–50 Schiele H, Calvi R, Gibbert M (2012) Customer attractiveness, supplier satisfaction and preferred customer status: introduction, definitions and an overarching framework. Ind Mark Manage 41(8):1178–1185. https://doi.org/10.1016/j.indmarman.2012.10.002 Schiele H, Horn P, Vos B (2011) Estimating cost-saving potential from international sourcing and other sourcing levers: relative importance and trade-offs. Int J Phys Distrib Logist Manag 41(3):315–336 Schuh C, Kromoser R, Strohmer M, Pérez R, Triplat A (2009) The purchasing chessboard: 64 methods to reduce cost and increase value with suppliers. Springer, Berlin, Germany Schumacher S, Schiele H, Contzen M, Zachau T (2008) Die 3 Faktoren des Einkaufs: Einkauf und Lieferanten strategisch positionieren. Wiley-VCH, Weinheim Steinle C, Schiele H (2008) Limits to global sourcing? Strategic consequences of dependency on international suppliers: cluster theory, resource-based view and case studies. J Purch Supply Manag 14(1):3–14 Steinle C, Schiele H, Ernst T (2014) Information asymmetries as antecedents of opportunism in buyer-supplier relationships: testing principal-agent theory. J Bus–Bus Mark 21(2):123–140 van Weele AJ (2005) Purchasing & supply chain management: analysis, strategy, planning and practice, 4th edn. Thomson Learning, London etc. Versteeg A (1999) Revolution im Einkauf: höchste Qualität und bester Service zum günstigsten Preis. Campus-Verl, Frankfurt/Main etc. Vos FG, Schiele H, Hüttinger L (2016) Supplier satisfaction: explanation and out-of-sample prediction. J Bus Res 69(10):4613–4623 Williamson OE (2008) Outsourcing: transaction cost economics and supply chain management*. J Supply Chain Manag 44(2):5–16

Chapter 5

Manufacturing Systems Henk Zijm

Abstract This chapter starts with a basic definition of manufacturing and a brief history on how the discipline developed throughout time, from initial craftsmanship via mass production to flexible and digital manufacturing. In this chapter, we use the term manufacturing to describe the process of converting raw materials or purchased components into physical products. We discuss basic performance measures to assess a manufacturing company’s efficiency and effectiveness. A major part of this chapter is devoted to a categorization of functions that together constitute an organizational framework for manufacturing, with a brief treatment of the essential elements of each function (advanced). Finally, we discuss the future of manufacturing by means of a short review of some important innovations induced by either technology progress or societal demands (state-of -the-art). We do not discuss prominent manufacturing planning and control systems, for this the reader is referred to Chap. 12 of this volume.

5.1 Manufacturing: Definition and Brief History Manufacturing is the process of converting raw materials or purchased components into physical products. It generally refers to all transformations that take place in a factory or plant and includes both machining and assembly operations, as well as internal transport. In almost all cases, the resulting products serve to fulfill a desired functionality that in turn determines how they are used. Products may be sold to consumers but may also serve as input for subsequent manufacturing or assembly processes; in many cases, complex final products are the result of a large sequence of manufacturing processes, performed in various stages and by different organizations in a supply chain or network. As an illustration, consider raw materials (e.g. iron ore) obtained through mining, which are converted into useable basic components (e.g. steel plates or castings) from which components are made by means of machining H. Zijm (B) Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, Overijssel, The Netherlands e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_5

75

76

H. Zijm

operations (engines and body parts for a car), which are finally used in a subsequent assembly process of a final product (a car). In this chapter, we restrict ourselves to manufacturing systems that take place at a single geographical location. Aspects of the broader concept of supply chains are discussed throughout this book, to start with the SCOR reference model presented in Chap.3. Manufacturing as a profession is almost as old as civilization. The production of pottery in ancient times, often decorated with paintings, can be seen as a manufacturing process, carried out mostly by skilled artisans. The Phoenicians were able to build seaworthy ships more than 2500 years ago, as did the Romans and the Vikings later. The cutting of stones and the building of the pyramids indicate high levels of craftsmanship and, not to be forgotten, work organization. The production of weapons, clothes and agricultural products again shows a long history of progress and refinement. Essentially however, all energy needed for the conversion processes was delivered by human and animal forces, without much mechanical aid (although the wheel and the lever can be seen as early tools to amplify human power). Indeed, the word “manufacture” stems from the Latin “manu factus”  “made by hand”. The first industrial revolution, lasting from about 1760 until 1820–1840, marked the start of manufacturing at an industrial scale, and hence meant the definitive transition from the classical domestic system and the craft guilds to mass production and mechanization. The invention of the steam engine by Thomas Newcomen, and its improvement by James Watt, but also the efficient application of water power and, based on these power generators, the development of machine tools were driving forces that enabled the birth of the factory system. Mass production requires physical concentration of production factors (capital, human labor, energy, machine tools and transport devices), and so the massive factories were born that colored the industrial landscape in the 19th and large parts of the 20th century. During the second industrial revolution, generally positioned between 1840 and 1870, steam-based transport (railways, inland vessels and ocean ships) and the machine tool industry developed further, enabling production and transport at a scale never witnessed before. The British textile industry was one of the early adopters of the factory system, while almost at the same time agricultural production started to benefit from mechanization. The substitution of coke for charcoal allowed the design of much larger blast furnaces for iron production, which in turn formed the basis of the massive steel industries that arose in the late 19th century. In the first decades of the 20th century, the assembly-line based manufacturing of automobiles in the Ford factories were a new milestone in industrial production. Scale, not scope, was the leading paradigm, as exemplified by the answer of Henry Ford to the question in what color the T-Ford was going to be produced: “Any color, as long as it’s black”. Mass production and limited product diversity continued to dominate industrial production, also during the first decades after the Second World War, to fill the shortages that resulted from the preceding time period. As prosperity started to grow from 1960 onwards, consumers began to demand larger variety, leading to more complex products. In response, manufacturing industries introduced versatile machines that were able to manufacture products in many variants, albeit often at the cost of large

5 Manufacturing Systems

77

changeover or setup times. Economies of scale and hence large batch production remained the leading philosophy. The two oil crises of 1973 and 1979 for the first time revealed the weaknesses of the prevalent manufacturing philosophy. Raw material prices and interest rates raised sharply and industrial companies experienced a phenomenon already observed by Harris in 1913: large batch production essentially leads to high inventories, not only of final products but also of intermediate materials and parts. Factories were able to produce efficiently, but could not cope with the flexibility that a changing society and changing markets required. In addition, publications such as “Limits to Growth” by the Club of Rome (Meadows et al. 1972) stressed the depletion of natural resources and the pollution of the natural environment at an exponentially increasing rate. For the first time industry and the public started to realize that current supply chains were to become economically prohibitive, and socially unacceptable. Another major problem concerned product quality, not so much with products delivered to the consumer but merely in the factories. Mass production without sufficient product and process quality assessment may lead to products that do not meet required standards in the final test, and hence are discarded. The result is a system that may be efficient but wastes a lot of materials. At the same time, Japanese manufacturers showed that efficient production of high quality products was possible without the burden of large stocks and wasted materials. The fact that Japan, more than any other country, lacked sufficient natural resources, may have helped leading production engineers to find unorthodox solutions when being confronted with the effects of the oil crises. But also the eastern way of viewing systems from a more holistic perspective as opposed to the fragmented western production philosophies helps to explain the success of Japanese manufacturing in the seventies and eighties of the 20th century. We return to this topic in Chap. 12. Gradually, it was recognized that next to efficiency and quality also flexibility, i.e. the ability to manufacture large varieties of high quality products in small batches is an essential performance criterion. Fortunately, new technologies proved to be at least a partial remedy. The introduction of flexible manufacturing systems, often based on computerized (CNC) machining and robotized assembly, helped to balance efficiency and flexibility. In addition, computerized information systems such as MRP and ERP helped to synchronize production across various stages in a way that was simply not possible without the computer. At the same time, attempts were made to simplify and synchronize the processes themselves, by adopting production philosophies such as Just-in-Time, or lean and agile manufacturing that focus on rigidly removing any buffer stocks as these were primarily seen as indications of waste or slack that characterize non-synchronized production. And now, when demand for more sustainable products and processes has entered the scene, we again need to rethink our manufacturing strategies.

78

H. Zijm

5.2 Manufacturing Systems: Fundamentals (Basic) The raison d’être of a manufacturing system is to make products that add value or that provide a desired functionality to their buyers. Hence, it is essential to know what potential customers value and how one should structure both the manufacturing processes as well as manufacturing logistics to optimally respond to these perceived market demands. Demand sensing and forecasting but also marketing are disciplines entitled to explore potential markets; aspects of these will be discussed in Chap. 6. Here, we concentrate on a characterization of both product/market combinations and manufacturing systems design that guide strategic choices on how to operate to respond to specific demand.

5.2.1 Product/Market Typology Knowledge of a company’s customer base for a particular set of products is essential to decide upon the organization of the necessary production processes. Product positioning strategies have been proposed by many authors, here we follow a characterization suggested by Fogarty et al. (1991), see also Zijm (2000). An important notion is that of the Customer Order Decoupling Point (CODP) which splits a chain of steps needed to manufacture a final product into forecast driven and customer order driven activities, c.f. Fig. 5.1. • Make and assemble to stock (MATS). This is the typical manufacturing philosophy for the majority of consumer products such as electronic equipment, food and drugs. In most cases, there is no direct relation between the manufacturer and the market; typically, the wholesaler and the retail sector are responsible for sales and services. Nevertheless, responsibility for a product’s performance resides with the manufacturer. The fact that legal obligations with respect to product safety have multiplied in the last decades may indeed force a company to initiate sometimes massive product call-back actions to not only prevent accidents but also to save its reputation. • Make to stock, assemble to order (MTS/ATO). When a large variety of different products is built up from a limited number of components, it makes sense to produce components to stock but to perform the final assembly based on customer orders (catalogue products). In this way, one avoids high final product inventories, but the price to be paid is of course the fact that customer service is no longer immediate. The manufacturing of cars and trucks is a good example of an MTS/ATO system. A fierce competition may force companies to reduce final assembly and delivery times as much as possible, often enabled by a far-reaching degree of automation in the final assembly process. • Make to order (MTO). Companies facing a high diversity of end-items in small quantities (small batch manufacturing) where the diversity originates already at the component level, typically operate in a Make-to-order mode. Most metal working

5 Manufacturing Systems

79

Fig. 5.1 Four CODP’s representing four product positioning strategies

(machine) factories belong to this category; their customers are typically OEM’s (Original Equipment Manufacturers) but also public utility organizations, not so much the end-consumer. Materials to be used are still universal and often procured based on forecasts. • Engineer to order (ETO). An engineer-to-order company typically designs and engineers products based upon a functional specification of the customer, and in close co-operation with the latter. Highly specialized equipment is typically produced in an ETO mode. Customers may be exploration industries in e.g. the oil and gas industry, ship owners and high-tech manufacturers. Only when agreement on the design is reached, the company starts to purchase materials, next manufactures parts and components and finally assembles, tests and installs the product. It should be said that, as a result of the thorough penetration of computer integrated manufacturing, but also through the introduction of product modularity, borders between some or the categories above are blurring. When building a product from standard modules, its final constitution may take place at the retailer’s shop or even at a customer’s home. The Swedish company IKEA is a fine example of the latter, offering an almost unlimited customer choice by designing standard modules for furniture that can be arbitrarily combined. The plug and play philosophy in the computer industry also is a good example of the same product philosophy (interestingly, however, a company like Apple was able to build an impressive market position, while not allowing its products to integrate with those of other computer manufacturers and software suppliers).

80

H. Zijm

A related phenomenon is that of mass-customization, i.e. the possibility to adapt a high volume consumer product to specific customer wishes. Examples can be observed in many sectors, ranging from household appliances to fashion. The basic strategy in these sectors is that of postponement, similar to the make to stock, assemble to order philosophy; producing a product or its constituting parts with all the functionality desired, but leaving it to the customer to select the final combination of parts, or to choose a personalized outlook (e.g. a print on a shirt, to be delivered by that customer). And finally, the advance of new production technologies may cause products to shift from one category to another one. One of the most prominent examples is additive manufacturing, better known as 3D printing, which is discussed in detail in Chap. 23. An early predecessor of 3D printing is stereo lithography which has been in use for more than 30 years as a technique for rapid prototyping but the current range of additive manufacturing techniques is much wider, and allows for the application of a much broader range of materials. It allows for the manufacturing of rather complex parts and products, but is still relatively expensive, needs much energy and is relatively slow, and therefore primarily suitable for low volume or one-of-a-kind manufacturing.

5.2.2 Manufacturing Process Typology A second major manufacturing typology pertains to the processes that are needed to build products. As with products, the borders between various types of manufacturing systems are blurring, primarily as a result of the deep penetration of automation and robotics on the manufacturing floor. Nevertheless, the following typology has proved to be useful when translating a manufacturing strategy into concrete production requirements. Continuous production. Continuous flow production is generally reserved for manufacturing in the process industries and for bulk materials. Oil refineries, the production of a wide range of chemical products but also food processing lines are examples of continuous (as opposed to discrete) production. Products are often liquids that can be packed in any amount desired; the packaging lines, although clearly ending with discrete products, are still generally considered as part of the continuous production system (e.g. a bottle filling line). Mixed model flow and assembly lines. This category encompasses many assembly processes as can be found in the automotive industry, in consumer electronics, as well as parts manufacturing systems that are based on a fixed, repetitive, sequence of process steps, basically identical for all products. In the past, often a distinction was made between dedicated and mixed model flow production, where the former refers to a line entirely dedicated to the manufacture of a single product. As a result of far reaching automation and robotics, but certainly also due to the rise of demand variability, dedicated flow lines hardly exist anymore (and are in fact unaffordable). Still, mixed model flow and assembly lines are typically suitable for the manufacture

5 Manufacturing Systems

81

or assembly of products that are sold in sufficiently large volumes, with limited basic product variety. Job shop manufacturing. Job shops are characterized by a highly functional process structure in which machines are grouped according to specific processes, such as milling, drilling, turning or grinding in a machine shop. Each product or small batch of products may have its own routing through the shop, hence the system can in principle handle a large variety of different products. Job shops are typically suited for small product quantities. Routes and machine processes in a job shop are typically determined in a preceding step, process planning, which also yields detailed machine instructions. In the final quarter of the 20th century, with the introduction of Computer Integrated Manufacturing Systems (CNC workstations, Flexible Manufacturing Cells) the ability to manufacture even larger product varieties increased due to the development of advanced NC-programming tools. Group-technology based or cellular manufacturing systems. In such systems, products are grouped based on similarity in production characteristics and consecutive processing steps will take place in the same cell (consisting of machines, tools and a small number of workers). Advantages typically stem from reduced materials handling and the fact that a team of workers is entirely responsible for, and very knowledgeable on, a specific set of products. As such, these systems form an interesting compromise between flow lines and job shops. On site manufacturing. Examples of on-site manufacturing include the realization of complex infrastructural works (bridges, tunnels) or the completion of a major industrial facility, works that are often organized as a separate project. These processes are characterized by the fact that equipment needed to realize the product is transferred to the product’s site, instead of the other way around, although a lot of components needed may be manufactured in a factory. Also here, one may observe a shift from one category to another one. In house building for instance, we observe the application of pre-fab constructions in particular in the lower price segments. An interesting other example concerns ship building where we may see the construction of modular ship segments at remote yards which subsequently are brought together, followed by assembly and further finishing operations on site.

5.2.3 Manufacturing Performance Measures The primary objective of a manufacturing organization is to ensure that the output of their processes meets market demand in volume, against agreed quality standards and at competitive prices. It goes without saying that the overall costs and benefits should leave room for a sufficient margin to guarantee continuity and to satisfy internal and external stakeholders (including but not limited to shareholders). To that end, a well-thought strategy and a close collaboration of marketing and sales, quality control, financial planning and control, human resource management and production and materials planning is essential. We will not treat all these functions in detail here but limit ourselves to a sketch of main performance criteria and an overview of

82

H. Zijm

organizational functions in the context of manufacturing management. In Chap. 12, we will discuss a number of Manufacturing Planning and Control Systems in more detail, while in Chap. 19 algorithms for decision support of these systems will be presented. As we already saw in the brief historical sketch at the beginning of this chapter, the objectives of a manufacturing company have shifted over time. Still, for many organizations, cost efficiency is a key performance index. Efficiency is a qualification on how much output is produced given the availability of certain amounts of resources (equipment, materials and manpower). The higher the output per unit of resource input, the more efficient the system is. Generally, we distinguish between technical efficiency (output quantities in relation to input quantities in a technical production sense) and economic efficiency (which is merely based on input and output prices). In the latter case, the cost to produce a given number of products should be as low as possible in terms of tariffs charged for the use of needed production factors (equipment, utilities, materials and manpower). As we will see later, this one-dimensional view on production costs may lead to significant problems at other parts in the supply chain. Already in the first decades of the preceding century, management started to realize that a strict focus on efficiency bears the risk of producing large quantities of products of which a significant portion turned out to be useless due to bad quality. That observation led to the inclusion of a second major objective: quality. Quality defines whether a produced product meets pre-defined technical and functional specifications. Statistical methods, as developed at Bell Telephone Company (Shewhart 1931), helped to systematically check product quality. Even more important was the early shift to methods of statistical process control, to check whether working methods and machine parameters are tuned such that bad quality production is prevented, and to adjust parameters in case of deviations. The focus on quality, advocated by industrial engineers like Edwards Deming (cf. Deming 1986), appeared to be one of the major factors behind the success of Japanese production systems, next to the third major objective: flexibility. Flexibility concerns the ease with which a production system changes between various product variants, or the ability to produce several variants of a product (almost) simultaneously, as well as its adaptability to (temporarily) changing volumes. The term flexible manufacturing systems is sometimes used explicitly to denote systems that consist of computer numerically controlled (CNC) machines, equipped with tool magazines and automatically tool-changing devices, together with automated material handling systems (AGV’s, pallet conveyor systems). However, flexibility is a much broader concept, sometimes the word agility is used to denote systems that are able to quickly adapt to changing requirements in terms of product volume and mix. The fourth and final objective often results once the first three objectives are fulfilled but nevertheless needs special mention: speed. Speed is a requirement put to many manufacturing systems and denotes how fast a system is able to respond to external demand. Clearly, fast delivery can also be

5 Manufacturing Systems

83

realized by producing items to stock from which demand is satisfied but for unique or highly specific products that is simple not an option. These four performance measures are all meant to audit the internal processes in a manufacturing company in the first place. Of course, they are implicitly reflecting also market demands (in particular quality and speed) but they are not explicitly dealing with the value that products and services of the company provide to the market. Indeed, for that reason, many authors have criticized the one-sided internal focus of these performance indices and have advocated a more holistic view on customer value management. We will return to the concept in several subsequent chapters.

5.3 Case Study: El-O-Matic The firm EL-O-Matic was founded in 1973 in the small town of Borne in the Netherlands and started with the design and sales of pneumatic aluminum actuators. In 1981 the company moved to the city of Hengelo. In 1990, it employed about 100 people in its main production facility in Hengelo, while small production facilities and/or sales offices were located in the UK, Germany, the United States and India. Starting with an annual sales of 11,000 pneumatic actuators in 1981, about 110,000 pneumatic and 10,000 electric actuators were sold in 1990. For 1995, a further increase to 177,000 pneumatic and 30,000 electric actuators was expected (of which 25% in the US and 10% in the Far East). As early as 1988, El-O-Matic was therefore rethinking its manufacturing strategy. Regarding its products, a product family structure could be discerned. ElO-Matic distinguished 12 basic types (sizes) of actuators. Each basic type can be delivered in a large number of variants. The most important parts of the actuator were the housing, two pistons, two end covers and the drive shaft. The diversity of actuators (the variants) is to a large extent determined by the actuator housings since these parts have to fit on the specific equipment of the customer. The conventional planning procedure operated as follows. Each month, a sales forecast covering the next 13 weeks (one quarter) was presented at product family level. A product family corresponds to the basic actuator types mentioned, while furthermore a distinction was made between products manufactured for the UK and for the Continent, leading to 24 different product families. The company needed a total lead time of 9 weeks, divided into three weeks for the procurement of raw materials (basic housings, shaft and covers, identical within one product family), two weeks for some basic parts manufacturing operations (variant independent), and four weeks for the variant-specific operations, some finishing operations and the assembly of the actuators. ElO-Matic promised customers a four-week delivery lead time (not including

84

H. Zijm

shipment and installation). The relatively low diversity during the first five weeks enabled the company to use forecasts to decide upon initial production quantities; only the last four weeks’ production (consisting of variant-specific operations) was based on customer orders. Facing the rapid product volume and mix increase in the late 1980s the company realized additional investments were needed in particular in metalcutting capacity, and after careful consideration decided to install a flexible manufacturing cell to perform a number of variant-specific processing steps on the actuator housings. Also, it experienced a severe pressure on its delivery lead times, and so the firm decided to install a Flexible Manufacturing System (FMS) for performing a number of variant-specific processing steps on the actuator housings. The FMS consisted of three CNC machining centers, linked by a pallet transport vehicle and an integrated pallet buffer system. Each machine was equipped with a tool magazine with an automatic tool changing device, while a tool robot provided the connection between a central tool store and the tool magazines. Housings were loaded and, upon completion of the FMS manufacturing cycle, unloaded again at an I/O station, equipped with special fixtures and clamping devices. The installation of the FMS enabled El-O-Matic to shorten their overall lead times drastically. The forecast-driven part of the production process still consisted of the procurement of raw materials and only a few pre-processing steps for which three weeks in total was reserved. The latter production phases (all metal-cutting operations, including those performed at the FMS), some finishing operations and final assembly took only three more weeks, yielding now a total lead time of six weeks. Since also the tapping of the screw thread is now performed at the MFS, together with all variant-specific operations, the number of basic housing types and therefore the number of product families reduced to 12. During the final stage, still some 150 variants were manufactured and assembled, based on customer orders.

5.4 Manufacturing Organization (Advanced) In Fig. 5.2 an overview is presented of the most important functions and their mutual relations in a manufacturing organization, ranging from long-term (on top) to shortterm (below), where we restrict ourselves to those functions that have a direct impact on manufacturing. For that reason, the marketing function is not separately discussed since in particular its branding function goes beyond manufacturing; aspects of marketing that directly influence demand are included in Long Range Sales Planning and Demand Management. The same holds for Human Resource Management, where

5 Manufacturing Systems

85

Fig. 5.2 Functions and their relations in a manufacturing organization

manpower planning aspects are included in Facilities and Resources Planning. Below, we discuss a number of functions in more detail. Long Range Forecasting and Sales Planning. Long range forecasting aims at the prediction of a market as a whole and is partly a method of expert judgement. A qualitative method to estimate future market volumes is e.g. the Delphi Method. With respect to quantitative forecasting methods, we distinguish between causal models and time series models. For the prediction of market volumes in the long run, causal models are usually exploited, for instance (multiple) regression models based upon earlier observed relations between various identified causal factors and realized sales volume for similar products (Makridakis et al. 1998). Subsequently, sales planning concerns the decision on a target market share for product ranges on a highly aggregate level, based on the market analysis mentioned above, as well as on an assessment of the power of competitors. Also, price setting constitutes an important instrument in gaining a particular market share; for that a well-thought and quickly manufacturable product design may result in a significant competitive advantages. Medium term sales planning in Make and Assemble to Stock (MATS) systems usually exploits time-based forecasting methods. This is quite different from the practice in Make to Order (MTO) and Engineer to Order (ETO) companies, where production plans need to be aligned with customer contracts that generally cover longer time periods (cf. Fig. 5.1). in addition, more and more MTO and ETO com-

86

H. Zijm

panies gain significant additional incomes from after-sales service contracts. In the long run, sales in an MTO or ETO environment heavily depends on specific customer relations for which general forecasting methods have less value. Primary contract winners are often those companies that are also frontrunners in technological product and process design. Product Design and Process Engineering. It is well-known that the far majority of costs made during manufacturing are determined by decisions made during the product and process design phase (Ulrich and Eppinger 1995). A well-known approach is for example Concurrent Engineering in which a new product range and the required processes are designed simultaneously, in order to prevent the design of superior products that eventually turn out to be too costly or, even worse, of which the time to design the corresponding manufacturing system, and hence the time to market, becomes too long because of engineering or quality problems in the start-up phase. A general guideline underlying Concurrent Engineering is also to postpone design decisions as long as possible, in order to increase flexibility during the corresponding technology selection processes. In addition, approaches such as Design for Manufacture and Design for Assembly have proven to be extremely valuable, cf. Boothroyd et al. (1994). Also, modularity of products and a high level of standardization of parts and components may help to reduce inventory investments considerably. Of the more technically oriented methods to quickly evaluate a proposed design, we mention rapid prototyping, see e.g. Kalpakjian (1992); interestingly, the most wellknown rapid prototyping technique, stereo lithography, is an early example of 3D printing, which recently has developed in a far more mature technology. Another important development concerns feature-based design; here features represent basic physical elements, i.e. a combination of material, physical shapes and tolerance measures. Various CAD (Computer Aided Design) systems exploit elementary and compound features as their basic building blocks (Rembold et al. 1993). Facility and Resources Planning. Both technological product/process design and commercial planning serve as input to the planning and possible acquisition of the facilities and resources that are needed. In this phase required resources, including manpower, machines and auxiliary equipment, are specified to enable the planned sales volumes to be realized. With respect to the selection of processes and equipment, decisions are fundamentally related to scale. If a product family is expected to run for a sufficiently long time in large volumes, it makes sense to install dedicated machines or e.g. a specialized assembly line. When volumes are only moderate, the product mix increases or product life cycles are relatively short, it makes sense to invest in more generic equipment. Most ETO and MTO systems exploit universal resources and make increasingly use of CNC workstations which can handle a large variety of (coded) work processes. Here, we also briefly pay attention to the selection of a process layout. Clearly, for high volumes to be assembled the classical (mixed-model) assembly line is a natural choice. On the other hand, the assembly of high capital goods (ships, aircraft) is usually done at site, with components and parts gradually moved to the spot. For parts manufacturing often a functional departmental structure is selected, in which case a smart layout planning is needed to minimize the costs of materials handling

5 Manufacturing Systems

87

between and within these departments (e.g. Francis et al. 1992; Tompkins et al. 1996). Still, in a rigid functional layout, transportation times and the sizes of batches may be significant, hence leading to high work-in-process inventories and overall long manufacturing lead times. For that reason, many authors are advocating a more product-focused layout, of which a cellular manufacturing system is a profound example (Wemmerlov and Hyer 1989). Burbidge (1990) suggests the application of production flow analysis as a means to arrive at a more group-technological process layout. Demand Management and Aggregate Capacity Planning. The implementation of this function clearly depends on the logistic product/market structure of the company. In medium to large volume manufacturing and assemble-to-stock products, the most wide-spread method is time-based forecasting of which the various versions of exponential smoothing are well-known and much applied examples. Trends and seasonal fluctuations are easily included in such methods. For a detailed account on timebased forecasting methods the reader is referred to Box and Jenkins (1970), see also Makridakis et al. (1998). Next, these demand forecasts are translated into prospective orders and finally order acceptance. Aggregate capacity planning in MATS systems involves the synchronization of production requirements with available resource capacities. Also the planning of additional shifts during certain periods and the decision to temporarily subcontract the production of certain components, may be a part of aggregate capacity planning, for which Linear Programming Models are an excellent tool, see e.g. Hax and Candea (1984), Winston (1994). Next, orders are accepted on a routine basis. In an MTO or ETO environment, the order acceptance function is generally more complicated, and includes the specification of functional and technical requirements, quality definitions, a delivery time and a price. Now, the role of aggregate capacity planning as a vital part of order acceptance is in particular to quote realistic customer order due dates, based on estimated internal lead times. In ETO companies, aggregate capacity planning does not only relate to the manufacturing divisions, but also to the design and engineering department. In all cases, a clear insight into the relations between the available resource capacities, a possible workload and the resulting manufacturing lead times is essential in order to determine sound inventory policies and to generate realistic customer order delivery times. We pay more attention to techniques for aggregate capacity planning in Chap. 19. Financial Management. Financial management in a broad sense regards all other functions but its main purpose is to keep the organization financially healthy (Wouters et al. 2012). Naturally, it assesses whether investments are allowable and determines the payback period (or break-even point) of any investment in equipment and resources. Another essential function is to provide the manufacturing organization with sufficient working capital (either internally or externally funded) to run the business smoothly. The amount of capital tied up in material and parts and products inventories limits the liquidity, i.e. working capital, of the organization, hence requires a close interaction with materials and inventory management. The same holds for purchasing; in particular long term contracts with suppliers should be approved by the finance department. Clearly, financial management is also in the

88

H. Zijm

lead when drawing up annual budget plans (look ahead) as well as the financial annual report (look back), and the interaction with external financial stakeholders such as banks, insurance companies and the external controllers. A topic that has received a lot of attention throughout the years is the determination of manufacturing costs. Obviously, costs of personnel and investments in machines, tools and infrastructure (i.e. both direct and indirect costs) are well-known but the annual costs clearly also depend on the depreciation rates and terms. Classically, the summation of these costs led for instance to a machine hour tariff and hence to the cost price of a product, but quite soon it was recognized that such a calculation does not provide a solid basis for e.g. make or buy decisions. Alternatives such as Activity Based Costing are better suitable to assess what activities contribute most to a product’s cost, or what are the main cost drivers. Based upon that analysis and on the company’s market position, sales prices are determined which form the basis for the company’s revenue as well as taxes to be paid. Process Planning. The function of Process Planning is to specify all the technical information needed before a production order, a job or part of a job can be executed. Usually we distinguish between macro and micro process planning, where the first concerns all decisions at a shop level, while the latter deals with the detailed machine and tool instructions. The way in which a production order is split into a number of potential production jobs, to be loaded on the various resource groups, is a macro-process planning decision, and the same holds for the specification of resource requirements (machines, operators and auxiliary equipment) and product routings. The determination of cutting patterns, and of tool speed and feed rates at machines are micro process planning decisions, captured in NC programs that subsequently instruct machine tools. Within an MATS environment, process planning is already performed during the process design phase and hence hardly plays a separate role. The same holds for an MTS/ATO system but for MTO and ETO environments the situation becomes quite different. In order to speed up the process planning activities, a selection is often made of possible processes, machining methods and tool combinations stored in a database, after which a CAPP (Computer Aided Process Planning) system automatically generates the NC programs. These CAPP systems in turn are often based on the use of process planning features (not to be confused with the design features discussed earlier) that specify basic material processing patterns (e.g. bending, material removal, welding patterns). The combination of many process planning features yields a complete machine instruction (Kusiak 1990). It is important to realize that in principle much freedom exists in the selection of machining methods and hence of routings and process plans. Currently, having more advanced CAPP systems available, process planning in many metal working factories is performed only a couple of days (and sometimes less) before actually processing a job. Consequently, it makes sense to take into account the actual work load on the shop floor when developing process plans for a new order, for instance with the aim to balance the load among various workstations. One way to significantly increase the loading flexibility on the shop floor is by developing several alternative job routings (cf. Zijm 1995).

5 Manufacturing Systems

89

Master Production Planning. Within the MRP terminology this function is often called Master Production Scheduling but it is important to note that this is only a schedule in time, which critically depends on the availability of all resources needed. A Master Production Plan is defined at end-item level for MATS systems but for MTS/ATO systems it generally operates at the level of parts or components, after which it is followed by a customer-order driven Final Assembly Schedule. To explain why that is needed, the concept of a Bill of Materials (product structure specification) is helpful, see also Orlicky (1975) and Chap. 12 of this volume. Each end-item demand is translated (exploded in the Material Requirements Planning terminology) via the Bill of Materials in dependent demand for lower level items that have to be produced in advance (using so-called off-set lead times), or should already be available in intermediate stocks. The problem however is that within an MTS/ATO system, customer order lead times are generally short, and hence the manufacturing of parts and components should be completed well in advance of knowing their exact demand. In particular when a large variety of end-items can be assembled from a relatively limited number of components, an MPP at end-item level would be unmanageable while an MPP at parts level, based upon aggregate demand, is perfectly useful. We will come back to Material Requirements Planning and alternatives in Chap. 12 in more detail. In Chap. 19, we discuss in detail the relation between resource capacity profiles and off-set lead times. Inventory Management and Materials Planning. Inventory management plays an essential role at both an aggregate and detailed level (Silver et al. 2017). When smoothing aggregate production plans, inventories naturally arise in case of a temporary foreseen shortage of resource capacity, making it necessary to manufacture some products or parts well in advance of their perceived demand. This is the capacity smoothing function of inventories. A second source of inventories is the production in batches, often due to reasons of economies of scale. As we will see in Chap. 12, the well-known Economic Production Quantity essentially balances the costs of the start of a production run (set-up costs) against inventory costs; if set-up costs are high it makes sense to produce a large batch which may be stored to gradually fulfill external demand while meanwhile the resources are used to manufacture other products. A third reason for holding inventories is to buffer against demand uncertainty or normal demand fluctuations; in order to be able to satisfy external demand, so-called safety stocks may be inevitable. Finally, obsolescence may occur, leading to excess part or product inventories, which may have to be disposed of or at best sold against discount prices. The Materials Planning function translates demand at MPP level towards lower level item demand, as discussed above, taking into account available inventories of lower level items. Materials Planning and Inventory Management together perform the materials supply function to each department or resource group, and hence represents an essential input to Job Planning and Resource Group Loading, as well as to Purchase Management. Job Planning and Resource Group Loading. Once customer orders or replenishment orders have been accepted and macro process plans have been determined, jobs can be constructed at the resource group level. Basically, a job can be seen as the

90

H. Zijm

restriction of an order to a specific department, work cell or resource group (often called production units, cf. Bertrand et al. 1990). However, several customer order related jobs may be combined into a composite job (batching) or one large job may be split into several smaller jobs, e.g. to balance the load among several work cells or to speed up work (lot splitting). In addition, the availability of parts or components in stock may also alter the lot size of a job (this is called netting in the Materials Requirements Planning terminology, see Chap. 12). It is important to notice that jobs are the operational entities to be controlled at the shop floor, starting with their release and ending with their formal completion. The simultaneous loading of various resource groups aims at matching the required and available capacity within each group, by considering effective resource group capacities as well as routing constraints of jobs between the groups, but without specifying in detail routing and precedence constraints of a job within a group. Planning is based on either customer order delivery dates, or inventory runout-times, and in turn defines internal release and due dates for each separate job. Purchase Management. This function takes care of the procurement of all components and materials that are purchased from external suppliers. It receives instructions from inventory management while the allocation of production jobs to time windows naturally depends on the availability of these externally procured materials. It is a strategically important function as well, typically the Purchase Department establishes contracts with a large number of external suppliers, specifies service level agreements in which supplier delivery lead times, fill rates and prices are set, and carefully watches delivery reliability (Schnor and Wallace 1986). It closely interacts with materials and inventory management and often has to act within strict budget limits. OEM’s of in particular capital intensive goods often primarily take the role of a system integrator with sometimes more than 60% of their total turnover consisting of purchased parts. An extensive treatment of purchasing and procurement is given in Chap. 4. Shop Floor Scheduling and Shop Floor Control. This is the level where the detailed scheduling of jobs on all workstations in a resource group takes place. The goal is to meet the internal due dates set at the higher Job Planning and Resource Group Loading level. Hence, at this level we typically deal with the sequencing of joboperations on individual workstations, but generally not with lot sizing aspects (these have been covered at the Job Planning level already). Shop Floor Control deals with the monitoring and diagnostics of all operations, reporting on quality aspects, and signaling major disruptions that may require a rescheduling or replanning phase. We pay more attention to shop floor scheduling in Chap. 19, see also Schutten (1998), or Pinedo and Chao (1999). Quality Control. Quality Control regards most processes at an operational level. First of all, it has an important role in checking incoming (purchased) goods on their conformance to agreed specifications. Such a check is mostly based on samples but a detection of non-conforming products may eventually lead to a full control. Today, many companies satisfy a range of ISO standards which has among others the advantage that a bad quality material or part is easily traced back to the time period it was produced and the machine that processed it, allowing the supplier to trace

5 Manufacturing Systems

91

other parts produced in the same batch. Naturally, Quality Control is also ultimately responsible for the quality of all products leaving the factory but many manufacturing systems also have in-line quality control checks in order to feed any problem back to the shop floor as early as possible. More general, since long a shift can be observed from product to process control, for instance by statistical process control techniques but today often by continuously monitoring machine and tool conditions (tolerances, stand time of tools), enabled by smart sensors, in order to prevent damage instead of repair afterwards (if possible at all). See Ishikawa and Lu (1985) or Taguchi (1986) for an overview of Quality Control philosophies and techniques. In the same spirit, we observe a shift from corrective via preventive to predictive maintenance, enabled by advanced condition monitoring techniques of machines and tools but also indicated by first signals of quality problems of parts or products delivered. This concludes the description of a manufacturing organization structure. We have decided to cover also a number of technological functions that are less often found in managerial textbooks. In the forthcoming manufacturing chapters we will primarily focus on the capacity planning and materials management functions at the tactical and operational level, not so much on the general organizational issues. In these chapters, we have deliberately decided not to separate material requirements planning and resource group loading, as is still done in many textbooks. We believe such a separation is not only artificial, but in fact the source of many problems. In Chap. 19, we describe how to integrate capacity and materials planning, based upon a workload control concept which highlights the tight relationship between effective capacity (throughput) and lead time management. First we take a look at current and future developments of manufacturing systems.

5.5 Future Manufacturing Systems (State-of -the-Art) After several decades of declining interest in manufacturing in both Europe and the US, its significance as one of the few ways to create wealth (Gershwin 1994) has been rediscovered. Both technological and societal developments are responsible for paradigm shifts in manufacturing system design and engineering. Below, we discuss the most important phenomena that have changed, or are currently changing, the landscape. New Materials and Manufacturing Technologies Research into new and lightweight (bio-)materials, polymer technology, bioengineering and nanotechnology has opened exciting possibilities to design an entire range of new products. These products find their applications in both complex industrial assets (e.g. lithographic systems to be used in new generations of semiconductor manufacturing equipment, membranes for separation at a molecular scale), medical instruments (e.g. lab-on-a-chip devices for quick infection or contagion detection) and personal life (e.g. ICT-inclusive wearables). Technologies like precision machining and additive manufacturing (3D-printing) are further steps towards

92

H. Zijm

mass-customization. 3D printing in particular is believed to have a major impact on small batch and one-of-a-kind manufacturing as well as in spare parts supply for maintenance, see Chap. 23 for a detailed discussion on its perceived merits. In all cases, the tuning of products to the specific need of the user is greatly facilitated by the versatility of 3D printing as an innovative manufacturing technology. Automation and Robotics, Internet of Things, Digital Manufacturing The application of robotics to replace or assist human labor in manufacturing and assembly has already been visible for a long time e.g. in automotive assembly lines, in precision machining, in Automatic Storage and Retrieval Systems (AS/RS) in warehouses, and more recently in the development of unmanned vehicles for both passenger and freight transport, and drones with so far primarily transport and surveying tasks. Their fast development has been made possible by the design of new generations of (micro-)sensors and actuators that enable high precision positioning of automatic devices. One step further is the Internet of Things in which, based on sensor information, devices automatically signal other devices that action has to be taken, e.g. to provide new materials or parts to the manufacturing floor, or to start a detailed equipment inspection after an indication of malfunctioning performance, based upon automatic condition monitoring. More general, engineers envision manufacturing systems in which both materials and machinery are able to communicate with each other and find solutions based on decentralized and autonomous decision making using state-of-the-art algorithms, often based on Artificial Intelligence. This opens the world of digital or smart manufacturing, sometimes also referred to as Industry 4.0, in which a high level of integration of functions of both resources and assets through automatic communication and actuation is expected. In Chap. 12 we pay more attention to recent developments in digital and cloud manufacturing and the advance of cyber physical systems. It goes without saying that the rapid advances in digital manufacturing require highly skilled workers trained in a variety of technical disciplines but moreover in interdisciplinary thinking. Already now, we observe a shortage of engineers in both basic and applied disciplines such as nanotechnology, bio-engineering, artificial intelligence to name but a few. To fill this gap is a major challenge for both industry and educational institutes at all levels alike. Circular Economy and Closed Loop Supply Chains, Sustainability Already since the first appearance of “Limits to Growth”, published by the Club of Rome (Meadows et al 1972) it is recognized (albeit not by everyone) that exponential growth may lead to the unavoidable depletion of natural resources. In addition, many industrial processes and products used have a profound negative impact on our natural environment, through emission of hazardous materials (CO2 , NOx , particulate matter), noise and stench, water pollution and infrastructural problems (e.g. congestion). The latter can be remedied by the design of cleaner engines, e.g. electric, hybrid or LNG-powered vehicles for city distribution and local passenger transport, but also through a better utilization of existing equipment by means of smart scheduling. To stop the depletion of natural resources requires a drastic change from a fossil-fuel based towards a renewable resources based economy, as exemplified by solar and

5 Manufacturing Systems

93

wind energy and in particular the re-use of materials or components from disposed end-of-life products. The emergence of the circular economy philosophy essentially describes a system in which waste is turned into fuel or resources for future production, as discussed e.g. by McDonough and Braungart (2002), and by an already longer existing research stream focusing on Closed Loop Supply Chains. The idea is simply to collect, or to set up a return flow process for, disposed products and assets that are disassembled after which components and/or materials are reused or recycled to basic materials for future use. Chapter 16 of this volume is entirely devoted to the design and analysis of Closed Loop Supply Chains. Reshoring Manufacturing, Social Responsibility From the sixth decennium of the preceding century onwards we have witnessed a growing disappearance of manufacturing and assembly to so-called low wages countries, in particular in the Far East (e.g. India, China, Korea) and more recently also to Eastern European countries. Initially, most outsourced production concerned the fabrication of low-value and often low-tech products. Increasingly, and as a result of massive investments in higher education, workers in the Far East have demonstrated to be able to manufacture high-tech products as well. Nevertheless, in the last decade we have witnessed a gradual reshoring of production back to Europe and the US. There are a number of reasons for that. First, wages in particular in Eastern-China have raised significantly compared to only 20 years ago. Second, the amount of manual labor of in particular high-tech products has decreased dramatically, making the low-wage advantage (if still existing) less dominant. Third: the logistics costs of transporting products overseas have increased significantly (partly due to high fuel prices but also due to international regulations regarding safety, security and environmental concern), making again far away production less attractive. And finally, there is a growing consciousness under both the consumer and industrial entrepreneurs of what is called Corporate Social Responsibility (partly as a result of stakeholder involvement), in favor of local-for-local production. Some authors use the word “glocal” as a shortcut for “global when needed, local when possible”. Although future supply chains are expected to remain global to a large extent, the trend to avoid useless or low value transport also from an environmental perspective will continue to help diminishing the ecological footprint of manufacturing and logistics.

5.6 Further Reading Some information in Sect. 5.1 is based on the first chapter of Hopp and Spearman (2000) who review the history of manufacturing, while in addition emphasizing the differences between production philosophies in East and West. A far more complete history is presented by Chandler (1977), although from a strong American point of view. Section 5.1 draws heavily on Zijm and Klumpp (2016) in which trends and developments in the broader field of supply chain management are reviewed. The product/market typology framework sketched in Sect. 5.2 follows Fogarty et al.

94

H. Zijm

(1991), see also Zijm (2000) and Silver et al. (2017). The case study in Sect. 5.3 stems from Zijm (1996). References for the various functions in a manufacturing organization are given throughout the text in Sect. 5.4 while furthermore Zijm (2000) served as a basis for this section. An excellent introduction to Manufacturing Planning and Control Systems is also provided by Vollmann et al. (1997). The relation between capacity and lead time management is discussed in detail in Hopp and Spearman (2000). Silver et al. (2017) provide an extensive overview of inventory management models and techniques. A must-read on the essence of Quality Control is Deming (1986). Further details on future manufacturing systems can be found in Chap. 12 which focuses on a manufacturing planning and control systems, including Enterprise Resource Planning, Just in Time Manufacturing, workload control and a further discussion on Digital Manufacturing.

References Bertrand JWM, Wortmann JC, Wijngaard J (1990) Production control: a structural and design oriented approach. Elsevier, Amsterdam Boothroyd G, Dewhurst W, Knight W (1994) Product design for manufacture and assembly. Marcel Dekker, New York Box GEP, Jenkins GM (1970) Time series analysis, forecasting and control. Holden-Day, San Francisco Burbidge JL (1990) Production control: a universal conceptual framework. Prod Plan Control 1(1):3–16 Chandler AD Jr (1977) The visible hand: the managerial revolution in American business. Belknap/Harvard University Press, Cambridge, MA Deming WE (1986) Out of the crisis. MIT Press, Cambridge, MA Fogarty DW, Blackstone JH, Hoffmann TR (1991) Production and inventory management, 2nd edn. South-Western Publishing Company, Cincinnati Francis RL, McGinnis LF Jr, White JA (1992) Facility layout and location: an analytical approach, 2nd edn. Prentice Hall, Englewood Cliffs, NJ Gershwin SB (1994) Manufacturing systems engineering. Prentice Hall Inc., Englewood Cliffs, NJ Hax AC, Candea D (1984) Production and inventory management. Prentice-Hall, Englewood Cliffs, NJ Hopp WJ, Spearman ML (2000) Factory physics: foundations of manufacturing management, 2nd edn. Irwin/McGraw-Hill, New York Ishikawa K, Lu D (1985) What is total quality control?. Prentice Hall Inc, Englewood Cliffs, NJ Kalpakjian S (1992) Manufacturing engineering and technology, 2nd edn. Addison-Wesley, Reading, MA Kusiak A (1990) Intelligent manufacturing systems. Prentice-Hall, Englewood Cliffs, NJ Makridakis S, Wheelwright SC, Hyndman RJ (1998) Forecasting: methods and applications, 3rd edn. Wiley, New York McDonough W, Braungart M (2002) Cradle to cradle: remaking the way we make things. North Point Press, New York Meadows DH, Meadows DL, Randers J, Behrens WW III (1972) The limits to growth. Universe Books, New York Orlicky J (1975) Material requirements planning: the new way of life in production and inventory management. McGraw-Hill, New York

5 Manufacturing Systems

95

Pinedo M, Chao X (1999) Operations scheduling with applications to manufacturing and services. Irwin/McGraw-Hill, Boston Rembold U, Nnaji BO, Storr A (1993) Computer integrated manufacturing and engineering. Addison-Wesley, Wokingham, UK Schnor JE, Wallace ThF (1986) High performance purchasing. Wight Publications Ltd, Essex Junction, VT, Oliver Schutten JMJ (1998) Practical job shop scheduling. Ann Oper Res 83:161–177 Shewhart WA (1931) Economic control of quality of manufactured product. D. Van Nostrand Company, New York Silver EA, Pyke DF, Thomas DJ (2017) Inventory and production management in supply chains. CRC Press (Taylor and Francis), Boca Raton Taguchi G (1986) Introduction to quality engineering—designing quality into products and processes. Asian Productivity Organization Tompkins JA, White JA, Bozer YA, Frazelle EH, Tanchoco JMA, Trevino J (1996) Facilities planning, 2nd edn. Wiley, New York Ulrich KT, Eppinger SD (1995) Product design and development. McGraw-Hill, New York Vollmann TE, Berry WL, Whybark DC (1997) Manufacturing planning and control systems, 4th edn. Irwin/McGraw-Hill, New York Wemmerlov U, Hyer NL (1989) Cellular manufacturing in the US industry: a survey of users. Int J Prod Res 27:1511–1530 Winston WL (1994) Operations research: algorithms and applications, 3rd edn. Duxbury Press, Belmont, CA Wouters M, Selto F, Hilton R, Maher M (2012) Cost Management: strategies for business decisions. McGraw-Hill, Berkshire Zijm WHM (1995) The integration of process planning and shop floor scheduling in small batch part manufacturing. Ann CIRP 44(1):429–432 Zijm WHM (1996) Shop floor planning and control for parts manufacturing shops. In: Van Beek P, Fortuin L, Van Wassenhove L (eds) OR at work. Taylor and Francis, pp. 111–132 Zijm WHM (2000) Towards intelligent manufacturing planning and control systems. OR Spectr 22:313–345 Zijm H, Klumpp M (2016) Logistics and supply chain management: developments and trends. In: Zijm H, Klumpp M, Clausen U, Ten Hompel M (eds) Logistics and supply chain innovation: bridging the gap between theory and practice. Lecture notes in logistics. Springer International Publishing, New York City, United States, pp 1–20

Chapter 6

Marketing Concepts and Instruments in Supply Chain Management Thomas Neukirchen, Oliver Gansser and Matthias Klumpp

Abstract Logistics and Supply Chain Management as a subfield of management science have their roots in the production and distribution of goods. In this chapter the basic objectives, principles and links about marketing and their relationship to logistics and supply chain management are outlined. The first—basic—part addresses the definition, logistics interaction, basic concepts, and a case study regarding marketing as a management philosophy and principle for corporate leadership. Here, “distribution channels” and “time to market” are two marketing topics particularly relevant to logistics. The advanced part of the chapter outlines detailed instruments for marketing strategies and market research. The final state of the art part of this chapter describes modern forecasting methods as well as innovation fields, including a further case study regarding future trends.

6.1 Definitions, Objectives and Logistics Interfaces (Basic) Modern marketing management is widely seen as a leading value driver for businesses around the world. Logistics and supply chain functions are important enablers of the value that consumers experience. Most product markets changed after the Second World War from a seller’s market to a buyer’s market as more and more suppliers offered an ever-larger volume and variety of products. In such situations, sales and earnings were not a given once a product has been manufactured. Saturation was T. Neukirchen (B) · O. Gansser FOM University of Applied Sciences, Essen, Nordrhein-Westfalen, Germany e-mail: [email protected] O. Gansser e-mail: [email protected] M. Klumpp FOM University of Applied Sciences and University of Duisburg-Essen, Essen, Nordrhein-Westfalen, Germany e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_6

97

98

T. Neukirchen et al.

characteristic for the market the companies had to face in these times. Competition among producing companies rose and customers could decide between variations of each product. Divided markets and smaller market shares for each product meant that it was more likely than ever for companies not to reach viable, that is, sufficiently high, sales volumes. Therefore, focus shifted towards the act of selling itself, the qualities of markets became the subject of professional inquiry. The professional management science discipline of marketing management came into being by providing elaborated methods and instruments for securing product sales. Obviously, the main concern of such strategies and activities are the characteristics, and (more or less stable or malleable) preferences of customers as in standard free markets they decide about such product sales (outside exceptional situations such as monopolies or government regulated markets). Modern marketing management has been defined as follows (e.g. Kotler and Keller 2015): Marketing management has the objective to professionally use company resources to increase the customer base, improve customer opinions of company products and/or services, and therefore enlarge the present and future earning potential. This includes the planning, implementation as well as tracking and review of all corporate marketing resources and activities. Furthermore, it aims at embedding the customer- and market-led management and decision philosophy strongly within all corporate processes and personnel on all hierarchy and decision levels.

In order to outline this marketing management perspective as the fundamental corporate orientation towards the customer and the market (including competitive forces such as technology, competitors, alliances, regulation and cross-market developments), we may look at the example of two companies and their market track record. In Table 6.1 we compare Amazon and Nokia regarding their strategic market orientation. Technology was used as primary enabler for both companies and they were able to provide customers with unique products or services of high value. Therefore they were global leaders in their domain. The difference is … Amazon still is an industry leader whereas Nokia is not. At one point in time, Nokia was not able to react adequately to the market needs at a strategic level, especially as smartphone applications began to dictate how mobile phones were used. This led a loss of one third of the global market share for Nokia between 2008 and 2011. On the other hand, Amazon is still able to maintain its leadership position in the online retail business, even entering new areas, e.g., grocery business. There may be changes ahead in technology, regional importance (Asia), political or other developments that may prohibit a successful company like Amazon to stay at the top of its game. Thus, companies need to understand changing customer demands and preferences and be able to meet their needs on a continuous basis. This is the final quest and challenge for marketing management: to secure future earnings potentials by scouting, analyzing, understanding and implementing the requirements of customers and markets, now and in the future. A good introductory example for the strategic nature of marketing is the PESTLE analysis explained next. Such an analysis with six dimensions is used for analyzing

6 Marketing Concepts and Instruments in Supply Chain Management

99

Table 6.1 Comparison of marketing success amazon and nokia Amazon Nokia Market Founding date

Retail 1994

Telecommunications 1967 (roots from 1865 in pulp and rubber production)

Location

Seattle, USA

Espoo, Finland

Development milestones

1995, first online book sale 1998, internationalization, e.g. Germany, Spain 2007, start of Amazon Prime

Employees • 2017 • 2007 • 1997 Turnover • 2017/2016 • 2007 • 1997 Profit • 2017/2016 • 2007 • 1997

• 306,800 • 17,000 • –150

1987, first mobile phone 1998–2011, global market leader mobile phones 2014, mobile phone division sold to Microsoft • 105,000 • 112,260 • 36,647

• 113,420 mil. US-$ • 14,840 mil. US-$ • 147.8 mil. US-$

• 23,600 mil. e • 51,060 mil. e • 9,700 mil. e

• 2,371 mil. US-$ • 476 mil. US-$ • −22.6 mil. US-$ (loss)

• −766 mil. e (loss) • 7,200 mil. e • 1.4 mil. e

Sources Internet sources (wikipedia, statista, wikinvest, nokia, amazon—January 2017)

potential market segments (in a qualitative way) before further scrutinizing and defining marketing strategies. This can be applied to business units or complete markets, e.g., when entering a new country for selling corporate products or services. Within the analysis, the abbreviation letters stand for: P for Political, E for Economic, S for Social, T for Technological, L for Legal and E for Environmental. Hence, an overall perspective of a market or segment environment from different angles is necessary to develop a sound marketing strategy. Several lead questions for implementing the analysis are suggested, for example: • What is the political situation and how can it affect the industry and sales? • What are important economic factors? • How much importance does culture have in the market and what are its determinants? • What technological innovations are likely to pop up and affect the market structure? • Are there any current legislations that regulate the industry or can there be any change in the legislations for the industry? • What are the environmental concerns for the industry? To understand the market, this framework represents one of the basic parts of strategic management that not only defines what a company should do, but also influences an organization’s goals and strategies. Obviously, the importance of each

100

T. Neukirchen et al.

factor may vary from market to market and country to country. It is similar to a SWOT (Strengths, Opportunities, Weaknesses, and Threats) analysis but in some dimensions more detailed. • Political: These factors determine the extent to which a government may influence the economy or a certain industry, e.g., by imposing a new tax or duty due to which entire revenue generating structures of organizations might change. Political factors include tax and fiscal policies, trade tariffs, and other factors that may affect the business environment (economic environment) significantly. • Economic: These factors are determinants of an economy’s performance that directly impact a company and have long term effects, e.g., a rise in inflation rate of an economy would affect the way companies’ price their products and services. Adding to that, it would affect the purchasing power of a consumer and change demand/supply models for that economy. Economic factors include inflation rate, interest rates, foreign exchange rates, economic growth patterns and others. It also accounts for the foreign direct investment (FDI) depending on specific industries as well as qualification and employment structures and qualities. • Social: These factors describe the social environment of the market, e.g., determinants such as cultural trends, demographics, and population trends. An example for this can be seasonal buying trends where there is high or low demand during holiday seasons. • Technological: Such factors relate to innovations in technology that affect operations of the industry and the market. This adheres to automation, research, and development and the level of technological awareness and advancement that a market possesses. • Legal: These factors have both external and internal aspects. Specific laws may affect the business environment in a certain country while there are certain policies that companies maintain for themselves. Legal analysis also takes into consideration both angles and then identifies the strategies in light of these legislations, e.g., consumer laws, safety standards, labor laws. • Environmental: Factors include those that influence or are determined by the surrounding environment. Factors of a business environmental analysis include but are not limited to climate, weather, geographical location, global changes in climate, and environmental offsets. The case study discussed in Sect. 6.2 shows how intertwined marketing management and logistics as well as supply chain management functions are. Like other complementary products and items such as bread and butter or cars and tires, marketing management and logistics management need each other and rely on each other for a successful joint market posture. There are many intertwined topics and processes between marketing management and additional corporate as well as supply chain functions. This section sketches the interesting segment of interactions from marketing management with logistics and operations processes, mainly in purchasing and manufacturing.1 1 See

also the previous Chaps. 4 and 5.

6 Marketing Concepts and Instruments in Supply Chain Management

101

Fig. 6.1 Marketing and supply chain interaction

Obviously, especially forecasting methods and their results directly impact the planning and decisions in purchasing and manufacturing: quantitative prognosis results for sales numbers and volumes as well as direct customer orders will directly translate into item purchasing volumes a company needs from suppliers as input to manufacturing processes. Also timing (which customer and market requires which product in what numbers at what specific time) is an essential input for purchasing, transportation and manufacturing planning, therefore not only sheer volume quantities are expected as results from the marketing management and forecasting methods. Figure 6.1 outlines this interaction, depicting the inherent marketing management circle in the upper part (analysis, strategy, forecasting, implementation, control). Subsequently, the forecasting function is a crucial impact for a multitude of supply chain, logistics and operations management processes: from purchasing (supplier selection, contracting and ordering) over inbound transport, manufacturing and added services (packaging and others) to distribution processes. Note that light/blue indicated functions are often performed by corporate partners in the supply chain such as logistics service providers (LSP). Modern supply chain and information system architectures are providing this core interaction with joint ERP and supply chain information systems, often implemented in internet- and cloud-based services and platforms. The detailed connection with the operational level in distribution is also outlined further in Chap. 14.

102

T. Neukirchen et al.

6.2 Case Study: Customer Orientation Repair Shops Background “Following a policy reform in the European Union, independent motor traders and service groups were granted better and fairer access to technical information, training, repair shop equipment, and original spare parts. Consequently, as market competition increased, the aftermarket operations of automotive manufacturers faced increasing pressure. To overcome new market threats, one of the world’s most successful manufacturers of premium passenger cars transformed their aftermarket logistics platform to improve service, enhance retention, and increase part sales. Customer Challenge With a distribution model configured around a single warehouse, the automotive manufacturer’s service centers received only one parts delivery each working day. Automotive parts were held in a centrally located warehouse where orders were taken until 18:00 each day for them to be delivered to service centers by 8:00 the following day. Because of this, whenever an additional problem was diagnosed in the service center, the end-customer had to make a return visit or the service center would hold the vehicle for an additional day for the required parts to arrive. Both scenarios, however, did not fulfill the levels of satisfaction demanded by its end-customers. In a bid to resolve this, the manufacturer redesigned its after-market logistics operations, developing a network of strategic local distribution centers that provided same-day parts deliveries. Supply Chain Solution Working in partnership with the manufacturer, DHL Supply Chain drew up a sustainable logistics solution that enabled 18 service centers across Scotland and Northern England to receive three additional part deliveries each day. Due to the widespread geographical coverage of service centers across the region, location was of primary importance. DHL Supply Chain had an established warehouse operation in central Scotland which was ideally positioned and had available storage capacity to accommodate the strategic stock-holding requirements. The warehouse was re-configured and 16,500 sq ft allocated to store the automotive parts. An optimized solution was also devised for receipt, storage, pick, and dispatch. With the warehouse operation responsible for fulfilling three daily order runs each day, three short peaks of activity occurred that aligned to the cut-off times for dealer dispatches. Activity spikes of this kind meant a dedicated team of full-time employees would not provide a cost-effective solution due to the periods of downtime between the highs of activity. A decision to create a multi-skilled workforce was made, enabling the automotive operation to draw on labor support from the existing operation whenever it was required. Selected employees underwent training and learned about the manufacturer’s

6 Marketing Concepts and Instruments in Supply Chain Management

103

processes and procedures for safely and accurately picking orders and preparing for distribution. The outbound transport element into the dealerships was awarded to a different third party logistics provider following a competitive tender. Customer Benefits The three-year implementation and contract phase went live in September 2010 and, after the first month of operations, performance indicators demonstrated the success of the start-up. The operation handled 2,035 inbound lines and made 530 dispatches to dealers, which translated into a 100% service level. Since establishing the local distribution, there has not only been a decrease in the number of return visits end-customers require but also in the amount of inventory held at service centers. In addition, overall parts sales has increased. After 18 months, the manufacturer recognized the outstanding achievements and awarded the contract “gold” status for service performance and greatly exceeding the inventory accuracy targets set. DHL is currently working with the premium car manufacturer to roll out and replicate the innovative aftermarket logistics model in key geographies around the globe. Underpinned by a culture of continuous improvement, SCM and logistics aims to drive efficiency improvements by sharing successful ideas across all operations—a testament to how its size and scale can be leveraged to deliver excellence worldwide.” (Source: www.dhl.com/content/dam/downloads/g0/logistics/case_stuies/dhl_ auto_same_day_parts_deliveries_case_study.pdf). Lessons Learned This case shows the inherent connection and collaboration of marketing and logistics efforts in order to increase customer service and satisfaction and finally realizing competitive advantage and earnings potential. The customer shall ideally be the major objective and driver of strategy and process changes within all corporate activities from purchasing to production, distribution and sales.

6.3 Basic Functions and Interactions: Markets, Distribution Channels and Partners, Time-to-Market (Basic) 6.3.1 Development of Marketing Concepts: The 4P Example Originally published by McCarthy, the 4P-framework for marketing decisions can be seen as synonymous to the marketing mix, which is one of the three best-known strategy-models (the other two being the 3C’s model and the Ansoff Matrix) and as

104

T. Neukirchen et al.

Product Price

Promotion

Place Fig. 6.2 4P model: marketing mix

such can be embedded into the numerous existing approaches to marketing strategy (others focusing on the relative timing of market entrance and entrant roles, diversification and integration, for instance). In the classic formulation, 4P represents the words product, price, promotion, and place. Each P is further defined and related to strategic decisions and actions. We give a short overview of the classic 4P and one variation with the Modern Marketing Management 4P as presented by Kotler and Keller (2015), depicted in Fig. 6.2. The classic 4P refers to “Product, Price, Place, Promotion”. Here, Product incorporates the variables product variety, quality, design, features, brand name, packaging, sizes, services, warranties, returns (Source: Kotler and Keller 2015). Price represents the list price, further: discounts, allowances, payment period, and credit terms. Promotion refers to sales promotion and sales force, advertising, public relations (PR), direct marketing. Under Place the classic model subsumes channels, coverage, assortments, locations, inventory, and transport. Kotler and Keller (2015) have offered a replacement model categorizing the marketing mix-components into “People, Processes, Programs, Performance”. Variations, up to 8 Ps and more, include partial to complete overlaps of these two. The broad scope and complexity of modern marketing and its treatments in applied and scholarly literature suggest a long tradition of marketing, approached systemically. While marketing practice has been around for ages, its theoretical counterpart however is fairly new and their interaction has undergone a number of paradigm shifts, or changes in focus. Table 6.2 presents a timeline of marketing thought.

6.3.2 Distribution Channels Two general marketing topics are of particular interest for logistics and supply chain management: “distribution channels” and “time to market”. Logistics has been described as standing at the intersection of purchasing and marketing, as is suggested with the place-category in the various multiple-P-frameworks. Making products available to customers in a way that is most efficient thus constitutes the main

6 Marketing Concepts and Instruments in Supply Chain Management

105

Table 6.2 Timeframe of marketing thought Paradigm: focus on … Description Production

Advent of mass-production Machine-made products are substituting manufactured goods (with manufactured in its literal meaning as ‘hand-made’). Focus on price, standardization in favor of customization: demand overhang

Product

Rising disposable household income and increasing market size. Attempts to raise output quantities while adding features: product complexity rises, as do prices

Sales

Supply overhang due to enlarged capacities. Supplier-centered approach on advertising and short term-oriented personal selling. Persuasion efforts to compensate for product shortcomings

Consumer

Aim: make products, thus organizations should fit customers’ needs/demands. The customer-centered approach entails the entire modern marketing concept as, for instance, structured with the 4P-model and its negotiation both with the outside environment and within the organization

goal of distribution. Distribution contributes to the product attributes, most simply by trade-offs such as speed versus upfront costs and waste. Generally, distribution and transport means have to be chosen according to end user needs and preferences. Distribution channels can therefore include long chains of intermediate actors whose roles in fact consist in cost reduction and increasing efficiency. Distribution takes on a very broad scope in a logistics context. Strategic matters occur all along the supply chain, starting at pre-production with procuring and transporting raw materials. All steps and movements related to intermediate products up to the completed product and its delivery may be included, whether it be on a business-to-business-level (B2B) or on a business-to-customer-level (B2C). This encompasses all related (strategic) decisions (e.g., store location, stock levels, and information systems). With a view to modes of transport and their selection, we encounter a problem based on several factors. Some factors are strategic, for instance, the choice of own transport method versus that of a competitors, product attributes, method choice made by competitors, costs associated with respective channel, reliability, time, security, traceability, and customer service level required (Coyle et al. 1988). As Fig. 6.3 shows, these categories translate into many quality determinants, which, in customer-centered marketing, amount to choice criteria. Of course, selection of transport modes is directly related to the choice of the distribution channel, which is the succession of intermediary agents a product passes, starting at the producing entity and ending at the consumer. Examples for intermediaries are wholesalers and retailers. Given the structure of the intermediaries and their interactions, we can further differentiate between conventional marketing channels, vertical marketing systems (VMS, i.e. corporate, administered, contractual VMS), horizontal marketing systems and multichannel marketing.

106

T. Neukirchen et al.

Fig. 6.3 Scores of importance and satisfaction assigned by shippers to twenty-three quality dimensions of rail shipments (Grue and Ludvigsen 2006)

6.3.3 Time to Market Time to market (TTM) describes two things: the time span from product conception to availability for sale (both requiring some specific definition), and the strategy that aims at optimizing this period. It is of special interest with respect to markets with high innovation frequency. Though suggestive in terms of measurability, practical definitions of both the starting and end point of this time span vary greatly depending on the type of industry, organization or the profession considered. Possible starting points are (for example): product idea, concept or draft approval, approval of financing/budget, actual placement of staff. Similarly, various definitions of the end include shipment or sales start date, a predefined sales figure or a well-defined transaction within the producing organization. Generally, these definitions are industry-specific. With respect to strategic considerations, mostly TTM-values achieved by companies from the same industry are relevant. Figure 6.4 depicts the cost implications of deviating from an ideal TTM. Given the product development process is sufficiently structured, no phase or step within the product development process can be skipped or moved to advantageous effect. Within a framework such as six sigma, time-saving-measures (in the short term) that override the given succession of procedures may cause a company to incur both higher costs and lower quality. In general, important variables shaping a TTM improvement efforts are innovation frequency of the industry, predictability of product release schedules, resource input goals and conditions (e.g., labor input/cost versus speed), adaptability, flexibility and desired responsiveness to mid-project requirement changes.

6 Marketing Concepts and Instruments in Supply Chain Management

107

Fig. 6.4 TTM-effects on cash flow (green-early to market, orange-late to market)

One instance from the area of software development shows how the introduction of agile methods has drastically reduced TTM. Because software products pervade the world’s functionality, placing focus on achieving higher quality can be leveraged with respect to many applications. Indirectly, TTM-reducing measures represent competitive challenges and opportunities in virtually all industries. Continuing performance management beginning with the initial project draft and accompanying all succeeding stages improves TTM as well as other critical aspects (e.g., defect avoidance or identification, customer service and acceptance rates, stability). Further, this illustrates the difficulties for clear definitions of the concept itself, because rapid prototyping, for instance, aims at delivering principally finished products or elements (Fig. 6.5). While waterfall methods make planning phases more stringent and apparently easier, numerous shortcomings, virtually with respect to all other areas, render them outdated in the presence of agile methods. Most important shortcomings are the separation of business and development in early phases, high knowledge requirements on the side of users (which is rarely given) and, in line with this, potential for extensive rework—all of this increasing TTM. With agile methods, flexibility, cost and risk reductions, short release circles (typically 2–4 weeks) greatly offset the more intense management required—and decisively reduce TTM.

6.3.4 Cows, Dogs, and Stars? Perhaps the most famous brand marketing tool has been known since the 1970s as the Product Portfolio Matrix, Boston Box, BCG-Matrix, or growth-share-matrix, as it basically plots business units or products to be analyzed along the dimensions (relative) market share and growth rate. The quadrants of the diagram (see Fig. 6.6) are dubbed cash cows, dogs, question marks, and stars:

108

T. Neukirchen et al.

Fig. 6.5 Waterfall versus agile software development

• Cash Cows: Located at the intersection of high market share and low growth, generate cash for the company’s operations. • Dogs: Units remaining at break even. Depending on perspective, these are valued more or less enthusiastically. If one views long term-oriented goals such as job and human capital preservation as part of a company’s tasks, these units are sustainable. If one views generating cash as the main objective, these units may be seen as indifferent. • Question Marks: Companies with a high market share and high growth potential either turn into dogs or stars, depending on performance. • Stars: These are startups which have succeeded to the forefront of a fast-growing market or—niche (compare Farris et al. 2010). The driving principle behind this is cash flow-management and the notion that cash would be generated proportionally to a product’s market share, while its usage is positively correlated to market growth rate.

6 Marketing Concepts and Instruments in Supply Chain Management

109

Fig. 6.6 BCG matrix

The Market Share is defined by Farris et al. (2010) as the ratio of a market accounted for by a specific (‘well-defined’) entity. unit mar ket shar ei 

unit salesi ∈ [0, 1] total mar ket unit sales

Multiplying this dimensionless value by 100 yields the percentage market share for the units sold by a company i. Adjustment for prices yields revenue market share: r evenue mar ket shar ei 

sales r evenuei total mar ket sales r evenue

Variations of these measures can be used, for instance, to express relative market share of brands against some competitor’s market share (Farris et al. 2010). The discrete growth rate g of a time-dependent value M (for example, market share) between two points in time, t and t 0 is defined as g

M(t) − M(t0 ) M(t0 )

and the continuous rate is given as w

dM 1 ∗ (t0 ) M(t0 ) dt

and the general discrete growth rate regarding a number of points in time as

110

T. Neukirchen et al.

 gr owth rate (t0 , t) 

M(t) M(t0 )

1/n − 1, with n  t − t0 .

A number of related concentration measures are used, with different definitions accounting for dispersion among firms’ shares. For reference, we list the Rosenbluth, comprehensive concentration, Linda, and U-indices as well as the Pareto slope (Curry and George 1983). A measure widely used in economic analysis and policy consulting is the Herfindahl index: H

N 

si2

i1

Here, si is the market share of firm i in the market considered.

6.4 Quantitative Methods (Advanced) As the plethora of marketing models already suggest, the field of marketing has long eluded the access of quantitative methods and still does so in many respects. This is due to numerous reasons on all levels and with respect to many interactions: complexity and nonlinearities. For example, the delayed, unsystematic or interrelated responses to marketing efforts, cannot be quantified satisfactorily. Measurements as well as modelling efforts suffer from the ineffectiveness of available research methods (for instance, questionnaires), interdependencies and rapid changes of variables. Notwithstanding these overall limitations, a respectable variety of methods exists. Each is applicable to some areas and questions in marketing. Still, use and fit of a method for a given question has to be assessed carefully. To relate the methods shown in this section to Logistics and SCM, one may think of how these affect decisions in purchasing and manufacturing. For example, sales volume forecasts for a manufacturing company translate into item purchasing volumes on the level of its immediate suppliers. This pattern may replicate multiple times all along the supply chain, downward to 2nd, 3rd, …, nth-tier suppliers. Also timing is critical for purchasing, transportation and planning. Thus, marketing management and forecasts are expected to yield sufficiently precise estimates with respect to volume and time. Below we replicate the taxonomy of the main quantitative methods in marketing by Moutinho and Meidan (2003). The fields sub-methods of which will be explained in more detail in the subsequent sections are indicated in Fig. 6.7. The characteristic feature of permanent reinvention may suggest itself to gametheoretic treatments, because this has a decisively strategic component. Considering game theory as a toolbox for modelling and predictive efforts concerning all sorts of competitive behavior renders it relevant to at least a few areas in marketing. In retail, for instance, market conditions are to be reasonably close to theoretical assumptions such as interactivity, complete and perfect information, dynamics and interdepen-

6 Marketing Concepts and Instruments in Supply Chain Management

111

Fig. 6.7 Marketing methods

dence, at least among the dimensions of pricing, stock determination, negotiations in general and advertising budget allocation. Because game theory begins with an axiomatic approach, thus restrictively defining properties of an agent, it contradicts the empirical approach marketing and market research take, as the following informal outline makes clear. The (classic) mathematical description of a game includes a finite set of players. Further, the following is assumed: each player has finitely many feasible choice options. These are called strategies and contain an exhaustive plan of action: Here, exhaustive means, ‘for any choice situation that is laid out in the game, no matter if it actually occurs. Players are assumed to have complete, ordered preferences over possible outcomes (often represented as von Neumann-Morgenstern utility functions). These assign real-valued utilities to each outcome of the game. Thus, its use is confined to decision-making vis-à-vis very well-defined problems. One could model a ‘supermarket game’ using a normal form setup (Jones 2003). In the example shown in Fig. 6.8 two supermarket chains may pick from three strategies (aggressive, moderate, passive). These strategies are aggregates of much more detailed choices each chain can make for promotions, price reductions, in store-advertising, and advertising in different types of media, for instance. Many more simplifying assumptions hold: the game can be seen as Chain I against all contenders in its market, represented as one Chain J. Each player knows about the strategies available to each other player. In Jones (2003) the example was used to illustrate an equilibrium point for the game. Each chain repeatedly engages in this competition, for instance on a weekly basis. In Fig. 6.8, the overall payoffs for mutual choice of passive strategies (20, 20) are highest, thus arguably being an incentive for collusion. However, this is not a stable situation, and thus no solution in terms of game theory. Here, one would argue that an equilibrium is reached once each player

112

T. Neukirchen et al.

Fig. 6.8 Retailing competition as a normal form game (Jones 2003, 58). Supermarket Chains I and J compete, whereas J can be seen as representing many individual competitors of I. Available strategies are aggregated as aggressive, moderate, passive. In each field, the first number represents payoff for Chain I, the second payoff for Chain J

has chosen a strategy and no player can benefit by changing strategies while the other players keep theirs unchanged (Nash 1950). As is indicated by the numbers in bold print in Fig. 6.8, unilaterally deviating from a passive or moderate strategy looks beneficial. With the assumed symmetry of information and each player doing the same, the game would reach stability only of both players act aggressively. Here, both (or all) chains are comparably worse off. Still, there is no incentive to individually change to another strategy. Formally, this follows from the assumptions. A real world-explanation is that promotions are manufacturer-driven thus as much harming profits of retailers as they are inevitable (Jone 2003). The scope of game theory in marketing research is thus quite limited. However, its principles can prove useful in the analysis of market entry and competitive/cooperative strategy (entry barriers and likely contenders in a prospective market, network effects, allocation of manufacturing tasks, information distribution to market, information acquisition behavior of customers, bargaining in general). Arriving at a crisp, purely quantitative formulation of objectives and measures seems rewarding. Many decisions related to marketing are too complex to be captured in an arrangement of a few elegant formulae. Other instances of purely quantitative attempts at incorporating the topic of marketing as a whole are physics-inspired, for instance (e.g. mimicking percolation to model adoption and spread of innovation). Thus, with the obvious need of addressing the complexity of marketing strategies, methods for market research purposes as well as for deriving sufficiently precise forecasts are explained in Sects. 6.5 and 6.6, respectively.

6.5 Market Research Methods Because organizations must make informed decisions, marketing managers must find out what they need to know to develop marketing objectives, select a target market, position (or reposition) their product, and develop product, price, promotion, and place strategies. Therefore, they need information. To make good decisions, marketing managers must have information that is accurate, up-to-date, and relevant.

6 Marketing Concepts and Instruments in Supply Chain Management

113

To understand these needs, they first need to conduct marketing research to identify them. Typically, marketing research is an ongoing process, a series of steps marketing managers take repeatedly to learn about the market. A company can carry out the research itself or commission another company to achieve the goal. Management must be informed to make decisions. A research process usually has seven phases (Solomon et al. 2017):

6.5.1 Steps in a Market Research Process a. Definition of the research problem. There are three questions to answer. • What is the research objective and what questions should the research answer? The strategic triangle can be used to analyze the situation if the problem is not directly apparent from the company. For example, it could be clear that the company’s revenues are declining. The analysis of the strengths and weaknesses and the coverage contribution analysis show, for example, that the sales program of the company is obsolete. This phase usually ends with the need for further information (through market research) to create a new concept with the content objectives, strategies and measures. The link between the strategic triangle (Fig. 6.9) and the concept pyramid consists of information on the three corner points of the strategic triangle and the analysis and forwarding of the relevant information for the creation of a new company concept (Gansser 2014a). • What is the total population of interest and what is the nature of this population? In the case of inadequate definition of population totals, systematically distorted selection of information subjects, or sloppy realization of the selection criteria, the results of the investigation may deviate significantly from the circumstances in the population of interest. This is to be avoided. • Which internal and external factors influence the current situation? The analysis of exogenous influences leads to an analysis of the environment, i.e., the chances and risks arising from the general environment (analysis of the general environment), the relevant markets, the sales markets, the procurement markets, the capital market and the labor market (market analysis), and the relevant economic branch as a whole. Internal factors as the cause of a crisis are generally limited to the way a company’s strategic business unit is viewed. In doing so, the performance-creating area is also affected by the area of corporate management (Gansser 2014a). b. Definition of research design. The research design concretizes what information marketers collect and what kind of study they perform. Research designs fall into categories of secondary research and primary research. Not all research problems need the same research techniques, and marketers solve many problems most effectively with a combination of different techniques. In a first step, market researchers always must ask whether the information they need to decide is

114

T. Neukirchen et al.

Fig. 6.9 Role of market research, company concept and situation analysis

already there. Data that are intended for a different purpose than the specific one, are called secondary data. Information that is sourced for specific problems is called primary data. Primary data includes demographic and psychological information about customers and prospects, customer attitudes, and opinions about products and competing products, as well as their awareness or knowledge of a product and their beliefs about the people using these products. c. Method for data collection of primary data. Primary data collection methods are described as either survey or observation. The degree of structuring is very high

6 Marketing Concepts and Instruments in Supply Chain Management

115

Fig. 6.10 Research designs (based on Solomon et al. 2017)

Fig. 6.11 Qualitative versus quantitative research

for quantitative data collection methods. Qualitative data collection techniques are not or only partly structured. It should be emphasized that the data collection tools are the same for both qualitative and quantitative research. Figure 6.10 illustrates this. Characteristically, three factors influence the quality of research results: validity, reliability, and representativeness. Validity is the extent to which research measures what it was intended to measure. Reliability is the extent to which research measurement techniques are free of errors. Representativeness is the extent to which consumers in a study are like a larger group in which the organization has an interest. It is about the predictability of the opportunity for all elements of the population to fall into the sample (Fig. 6.11).

116

T. Neukirchen et al.

d. Design of the Sample. Market researchers usually collect most of their data from a sample of the population of interest. Based on the answers from this sample, they infer these results on the population. Whether such conclusions are true or inaccurate depends on the nature and quality of the sample. There are two main types of samples: Probability and Non-Probability samples. A probability sample is a sample in which each member of the population has a known chance of being included. A nonprobability sample is a sample in which personal judgment is used to select respondents and a convenience sample is a nonprobability sample composed of individuals who just happen to be available when and where the data are being collected. e. Data collection. Primary data can be collected in many ways. Surveys, physical measurements and observation are three of the possibilities. The quality of our inference is only as good as the data are. The same logic applies to the people who collect the data: the quality of the research results is only as good as the worst interviewer in the study. Therefore, interviewers must be trained and cared for. f. Data analysis and data interpretation. The basic prerequisite for data analysis is the determination of the measuring level of the available data. As a rule, the application of particularly powerful methods of statistics is only permitted if certain measurement levels are available. Measurement is the assignment of numbers or other symbols to characteristics of objects according to certain prespecified rules. Scaling is the generation of a continuum upon which measured objects are located. There are four primary scales of measurement: nominal, ordinal, interval and ratio. A nominal scale is a scale where numbers serve only as labels or tags for identifying and classifying objects with a strict one-to-one correspondence between the numbers and the objects. Ordinal scale is a ranking scale in which numbers are assigned to objects to indicate the relative extent to which some characteristic is possessed. Thus, it is possible to determine whether an object has more or less of a characteristic than some other object. Interval scale is a scale in which the numbers are used to rank objects such that numerically equal distances on the scale represent equal distances in the characteristic being measured. Ratio scale is the highest scale. This scale allows the researcher to identify or classify objects, rank order the objects, and compare intervals or differences. It is also meaningful to compute ratios of scale values (Malhotra 2009). Quality of the data: This refers to how normal the data is distributed. The first techniques discussed are sensitive to the linearity, normality and the same variability assumptions of the data. Studies of distribution, skewness and curiosity are helpful in the examination of the distribution. It is also important to understand the size of the missing values in observations and to determine whether to ignore them or to refer values to the missing observations. Another data quality measure is outliers, and it is important to determine whether the outliers are to be shifted again. When held, they can cause a distortion of the data; when they are eliminated, they can help with the assumptions of normality. The key is to try to understand what the outliers represent (Diez et al. 2014).

6 Marketing Concepts and Instruments in Supply Chain Management

117

Statistical methods can be classified as follows: • Descriptive data analysis: One of the tasks of statistical methods is to summarize data on many individual cases (for example: consumers, companies). Statistical metrics and representations are used in the form of tables and graphs. • Multivariate analytical methods: In market research marketers must deal with complex relationships between numerous variables. For instance, aspects of consumer behavior (e.g. brand selection, type of needs) can hardly be explained by one variable, and the success or failure of a product never depends on one factor (e.g. advertising budget or price). For marketers, therefore, multivariate analytical methods, which are suitable for the simultaneous analysis of many variables at the same time, play an important role (Malhotra 2009). a. Dependency analyses: There are procedures which are designed to explain a dependent variable by a certain number of independent variables, for example the market share of a product by advertising budgets, price, purchasing power of the target group, relative product quality, etc. Common methods are analysis of variance, regression and conjoint measurement. b. Interdependency analyses: In other multivariate procedures, connections between a larger number of variables are the focus. The variables are not classified as dependent or independent; rather, the whole set of interdependent relationships is examined. Common methods are principal component analysis, exploratory factor analysis and cluster analysis. g. Report the Research Results. The last step in market research is to report on the research results and to document the results. In general, a research report must be clear and concise. The readers—top management, clients, creative departments and many others—must be able to easily understand the results of the research.

6.5.2 Multivariate Analysis Methods in Market Research The purpose of this chapter is to provide a comprehensive understanding of common multivariate analytical methods in market research, leading to an understanding of the corresponding use of each of the techniques. We do not discuss the underlying statistics of each method. Rather, it is a field leader to understand the types of research questions that can be formulated, and the skills and limitations of any technique in answering these questions. In this section, the essential application of the method of market research like regression analysis, logistic regression analysis, factor analysis, structure equilibrium models, conjoint analysis and cluster analysis are described.

6.5.2.1

Analysis of Variances

The analysis of variance (ANOVA) can be used to check whether there is a significant difference between two or more mean values. The variance analysis is particularly

118

T. Neukirchen et al.

suitable for comparison between groups, which in turn explains their application for the evaluation of experiments, where comparisons between measured values from experimental and control groups must be made. The number of groups (independent variable) determines whether it is a one-factor analysis (one group) or a multi-factorial analysis (several groups). In variance analysis, a distinction is made between the explained and unexplained variance of the dependent (metric) variables. The influence of the independent variables (group membership) is assessed based on the relation between explained variance and unexplained variance. One of the central ideas of the variance analysis is to compare variances of the dependent variables within the groups with variances between the groups (deviations of the group mean values from the total mean value). If the variance between the groups is large compared to the variance within the groups, then this indicates a clear influence of the independent variables, which determines the group membership. Assumptions of the variance analysis (Diez et al. 2014): • The error term must be normally distributed, which at the same time is a normal distribution of the measured values in the population. • The error term must be the same or homogeneous between the groups. • The measured values must be independent of each other.

6.5.2.2

Regression

The central idea of this method is that the different values of a dependent variable (target variable) are to be fed back to another (independent) variable (influencing variable). In this sense, the dependent variable is explained by the independent or explanatory variables. The regression analysis method for analyzing associative relationships between a metric-dependent variable (target variable) and one or more independent variables (influencing variable) can be used in the following ways: • To determine whether the independent variables explain a significant variation in the dependent variable: whether a relationship exists. • To determine how much of the variation in the dependent variable can be explained by the independent variables: strength of the relationship. • To determine the structure or form of the relationship: the mathematical equation relating the independent and dependent variables. • To predict the values of the dependent variable. • To control for other independent variables when evaluating the contributions of a specific variable or set of variables. Steps of Regression Analysis (Diez et al 2014; Chapman and Feit 2015; Malhotra 2009): 1. Formulation of the regression model: Based on theoretical and empirical findings as well as previous experience, it is necessary to determine which independent variables could explain the variable of interest (dependent).

6 Marketing Concepts and Instruments in Supply Chain Management

119

2. Estimation of the parameters of the regression model: For the estimation of the regression model, some assumptions are necessary, which can be inferred from the relevant literature. A regression model in the bivariate case looks like Yˆ  b0 + b1 ∗ X and in the multiple (multivariate) case Yˆ  b0 + b1 ∗ X 1 + b2 ∗ X 2 + bn ∗ X n where Yˆ is the dependent variable, X 1 , . . . , X n are independent variable, b0 is the intercept and b1 , . . . , bn are the slopes of the corresponding variables. The most commonly used technique for fitting a function is a minimum square estimate (least squares procedure). This technique determines the best-fitting by minimizing the sum of the squared vertical distances of all the observations. The determined parameters (regression coefficients) determine the relationship between the independent and the dependent variables for the examined data record. By using these parameters and the respective variable values of the independent variables, the value of the dependent variables can be estimated for each case. 3. Checking model fit: An important measure for the assessment of a regression model is the proportion of the variance explained by the model divided by the total variance. The strength of association is measured by the square of the multiple correlation coefficient R2 , which is also called the coefficient of determination. R2 is between 0 and 1. The extreme values 0 and 1 mean that a model does not explain any variance or the dependent variable is completely explained.

6.5.2.3

Logistic Regression

With logistic regression, the influences on a dependent nominal scaled variable can be examined. It is assumed that the dependent variable is dichotomous, that means, it can take only two values (0 and 1). With the help of several independent variables, the probabilities for the values of the dependent variable (an “event”) are estimated. In a multinomial logistic regression, categorical dependent variables with more than two expressions can also be analyzed. Because no linear relationship is tested, the Regression coefficients are not interpreted exactly as in linear regressions. Only the direction of the influence can be interpreted.

6.5.2.4

Dimension Reduction with PCA and EFA

Data often have many variables—or dimensions—and it is beneficial to reduce them to a smaller number of variables (or dimensions). Contexts between constructs can be

120

T. Neukirchen et al.

identified more clearly. There are two common methods to reduce the complexity of multivariate metric data by reducing the number of dimensions in the data (Chapmann and Feit 2014): • The principal component analysis (PCA) attempts to find uncorrelated linear combinations that capture the maximum variance in the data. The direction of view is from the data to the components. • The exploratory factor analysis (EFA) attempts to model the variance based on a small number of dimensions, while at the same time trying to make the dimensions of the original variables interpretable. It is assumed that the data correspond to a factor model. The direction of the view is from the factors to the data. Reasons for the need for data reduction: • In the technical sense of dimensional reduction, we can use the factor/component values instead of variable sets (e.g. for mean value comparisons, regression analysis and cluster analysis). • We can reduce uncertainty. If we believe that a construct is not clearly measurable, the uncertainty can be reduced with a variable set. • We can simplify the data acquisition effort by focusing on variables that are known to make a significant contribution to the factor/component of interest. If we find that some variables are not important for a factor, we can eliminate them from the record.

6.5.2.5

Multidimensional Scaling (MDS)

MDS is a method that can also be used to find low-dimensional representations of data. Instead of extracting components or latent factors, as with the PCA or EFA, the MDS instead works with distances (or similarities). The MDS tries to find a low-dimensional map that best preserves all observed similarities between objects. Information provided by MDS has been used for a variety of marketing applications (Malhotra 2009): image measurement, market segmentation, new product development, assessment of advertising effectiveness, pricing analysis, channel decisions and attitude scale construction. In a study to explore behavioral types by the FOM in 2016, 22.131 people were asked on a scale from 1 to 7 about their values and their purchasing behavior. With PCA thirteen principal components were formed (five dimensions of human values and eight dimensions of consumer behavior): • Human values are: Enjoyment, Appreciation, Conformism, Security and Consciousness. • Consumer behavior are: Perfectionism, Brand Consciousness, Novelty Fashion, Stress, Price, Impulsion, Confusion and Habit. To represent these thirteen dimensions in a two-dimensional space, a metric MDS is calculated. To calculate the MDS, the individual Dimensions are correlated with each other (product-moment correlation). With this approach, the dimensions are

6 Marketing Concepts and Instruments in Supply Chain Management

121

Metric MDS with 13 Dimensions

mds[.2]

–0.6 –0.4 –0.2 0.0

0.2

0.4

0.6

Consciousness Security

Novelty Fashion

Enjoyment Perfectionistic

Conformism

Price Appreciation

Habitual Confused

Brand Conscious Impulsive

Stress

–0.5

0.0

0.5

mds[.1]

Fig. 6.12 Perception space; 13 dimensions, human values and consumer behavior

represented as points in the multidimensional space so that the distances between the points represent the correlations of the dimensions (Fig. 6.12). For non-metric data such as rankings or categorical variables, an MDS algorithm is used which does not take any metric distances (Chapman and Feit 2014). For practical application, it is important to get a plausible interpretation of the perception space created by the MDS. In this context, the additional integration of independent assessment dimensions into the perception space with the aid of the vector model can represent a valuable interpretation aid. In a pre-study of the behavioral types in 2014 (n  15.563), 40 values were calculated by MDS in a perception space. For interpretation, feature vectors in the form of the dimensions of the purchasing behavior are now included in the configuration of the MDS analysis (Fig. 6.13). The method implementation process can be described in short as follows: • First, the six dimensions of purchasing behavior were correlated with the individual items of the value orientations. • Subsequently, a linear regression analysis was carried out for each of the dimensions of the purchasing behavior. • The two coordinates of the values in the MDS are the independent variables which are used to explain the variance of the purchasing behavior (here the correlation with the values). • For the vector model, the beta coefficients of the two dimensions are of interest. These are used as coordinates for the vector to be drawn. • The vector runs in the diagram as a straight line through the origin and the point defined by these two coordinates, namely as an arrow in the direction of the point.

122

T. Neukirchen et al.

Fig. 6.13 Perception space with an integrated vector model (Gansser 2014b)

6.5.2.6

Cluster Analysis

Cluster analysis is used to find homogeneous groups (usually observations) within the data which are as heterogeneous as possible between the groups. Market segmentation is a typical application of cluster analysis. To determine the similarity of observations, different distance measures can be used. For metric features, for example, the Euclidean metric is often used, that is, similarity and distance are determined based on the Euclidean distance. Other measures like the Manhattan or the Gower distance are also possible. These have the advantage that they can be used not only for metric data but also for mixed variable types. Typical steps of cluster analysis (Chapmann and Feit 2014): • Selection of the variables to be used for group formation (e.g., sociodemographic features, setting variables, lifestyle characteristics). • Quantification of similarities or dissimilarities of objects based on a so-called proximity measure and determination of a distance or similarity matrix. • Summary of the objects into homogeneous groups based on the values of the degree of proximity using the application of a fusion algorithm (Fig. 6.14).

6 Marketing Concepts and Instruments in Supply Chain Management

123

Fig. 6.14 Cluster plot for the seven behavioral types

6.6 Forecasting Methods (State of the Art) Among the large number of forecasting instruments in financial management and controlling, only the three most important methods for predicting the behavior of market participants are presented—Conjoint analysis, marketing intelligence and Monte-Carlo simulation.

6.6.1 Conjoint Analysis Conjoint analysis is a widely used and established method for measuring preferences. In practice, they are mainly used for price estimation, new product planning, and for customer segmentation. In comparison with other methods, conjoint measurements are a more realistic form of preference measurement with a higher validity. Depending on the procedure, one or several product concepts are submitted for assessment. The products are defined by features that have a certain set of characteristics. Thus, the subject identifies shared values for each characteristic of a feature. Based on the measured preferences, a prognosis can be generated, which product is preferred and likely to be bought in the future. Conjoint analysis has been used in marketing for a variety of purposes, including the following (Malhotra 2009):

124

• • • •

T. Neukirchen et al.

Determining the relative importance of attributes in the consumer choice process. Estimating the market share of brands that differ in attribute levels. Determining the composition of the most preferred brand. Segmenting the market based on similarity of preferences for attribute levels.

To calculate these predictions, purchase decision models are applied. For various hypothetical product alternatives, total utility values are estimated, which are subsequently transformed into selection probabilities. For all models, the choice is based on the principles of utility maximization, so that alternatives with higher benefits are preferred to alternatives with lower benefits. The use of such decision-making models for forecasting purposes is problematic if there is no information on the real purchasing decision processes and therefore the market researcher must make an individual selection of the decision models. This disadvantage exists in the group of traditional conjoint analysis methods, in which the assessed alternatives are placed in a preference ranking of the information subjects or are evaluated by means of rating scales. This disadvantage is eliminated in the choice-based conjoint analysis, in which the persons select the most attractive alternative from different choice sets. A choice set consists of two hypothetical alternatives and the possibility of non-selection. Thus, it can be assumed that the natural buying behavior of the person is analyzed. The choice-based conjoint analysis is a probabilistic method for the preference structure measurement. The partial utilities of the individual characteristics are estimated from the total benefit. The assessment objects are constructed based on experimental designs. Procedure for conjoint analysis (Gansser and Füller 2015) • • • • •

Selection of the characteristic expressions Definition of experimental design Creation of orthogonally fractionated choice sets Presentation of the stimuli from the survey participants Estimation of the utility function.

The result of the conjoint analysis is the calculation of odds ratios. In the case of nominal characteristics, the odds ratio (exp(coef)) of a variable indicates the chances of the odds of a characteristic of a feature compared to the basic category, i.e., the ratio of a chance. It is then possible to interpret the feature expression with the highest chance ratio as an example for each characteristic compared to its basic category. Finally, the conjoint analysis can be used to answer which offer persons prefer and are likely to buy in the future. In addition to the chances of a feature compared to a basic form, this method also allows the relevance and therefore the importance of different characteristics to be measured.

6 Marketing Concepts and Instruments in Supply Chain Management

125

6.6.2 Marketing Intelligence A marketing intelligence system includes a set of procedures to maintain everyday information about developments in the marketing environment. The purpose of the information collection is the accurate and confident decision-making in determining the marketing concept (goals, strategies and activities). Once the data are collected (manually or automatically), the analysis is usually carried out using software-based systems. The intelligence approach is that the sources of information are of a different nature and are placed in a single environment after being captured. The goal is the integrated information collection and visualization of internal and external data sources. This enables current key performance indicators (KPI’s) to be viewed in real-time, or as fast as the data can be captured, and to analyze trends. The term “business intelligence” (BI) has become established as a concept for all methods for analyzing business performance. In this way, the company has different areas of the intelligence approach, which are aimed at analyzing and optimizing the partial performance. In addition to marketing intelligence, the sales intelligence division has also become established. In both methods, the requirement for the integrated efficiency measurement across the departments consists. The trend is that classic controlling tasks are transferred from the central control department to the operating divisions. Business Intelligence (BI), as a closed system, includes all the analysis and optimization capabilities that can be used to capture, analyze, and improve business information. Analyses in the context of intelligence approaches should meet the following requirements: • Evaluations must be possible in real time. • They should be accessible inside and outside the company with web applications (also mobile). • The data are available in standardized and consolidated form. They do not have to be collected, consolidated and evaluated by the user. • Users can customize the analyses and reports to their individual requirements. • For the analyses, menus are available for clear and meaningful charts. • Multi-dimensional analyses (OLAP) and data integration from all business areas allow to generate new findings. To efficiently capture data from the environment and the strategic triangle, some specific data sources are being developed that are particularly suitable for marketing intelligence. These data can usually be collected by the company itself. In the case of less sensitive data, there is also the possibility of commissioning external agencies: • The sales assistant as a free market researcher: Field service persons are generally closest to the customer. They can observe the way customers use the products most easily and without complex market research. Ideas for new products can be generated in this way. In addition, information about competitors and retailers can be recorded.

126

T. Neukirchen et al.

• Mystery Shopping in the retail store: Using camouflaged employees and their observations, the quality of consulting and competence of the salesmen should be determined. This is not undisputed and should be provided with an extended focus also on the quality of the facilities. Companies can also assess the quality of customer experience with the use of Mystery Shoppers. • Competition analysis: This can be done by purchasing the competitor’s products, reviewing the advertising campaigns, press reporting, reading their published reports, and so on. Competitive intelligence must be legal and ethical. • Customer community: Customers (size, demand, representativeness) that are to be identified can provide valuable information on the product, product use and sales channels as participants in a community (online or offline) or a panel. This enables them to be actively involved in the company’s improvement processes. Online platforms such as chatrooms, blogs, discussion forums, customer review boards can be used to generate customer feedback. This allows the company to understand customer experiences and impressions. • Official data: Governments from almost all countries publish reports on population development, demographic characteristics, agricultural production and other data. These country-specific basic data can be helpful when planning business concepts.

6.6.3 Risk Analysis with Monte-Carlo Simulation Marketing planning attempts to capture the future market share. Based on assumptions, future values are determined and target values are calculated. For the calculation of the target variable, a model is formulated with cause-effect relationships. Because the target variables are indeterminate, various risk scenarios are identified in a risk analysis by means of a Monte Carlo simulation. By means of Monte Carlo simulation, realizations of each influencing variable are generated according to a presumed probability distribution, from which values of the target variable are calculated, Based upon a large number of simulations, we hence obtain estimations of the probability distribution or at least mean, variance and confidence intervals.

6.7 Case Study: Telecommunication Customer Segmentation Using Machine Learning Background With rising consumption of data-intensive mobile content, a number of challenges for telecom providers arises: a search for better rates, signup benefits or discounts, wireless local number portability and business models relying on separation of mobile tariff and handset, increase of the number of customers willing to churn more frequently, etc.

6 Marketing Concepts and Instruments in Supply Chain Management

127

Customer Challenge The Irish company iD Mobile Ireland approaches these market characteristics by separating mobile tariff and handset, thus allowing customers to buy a new handset every three months given their current has been paid for, and further enabling customers to adjust the limits for calls and data services in their contracts on a monthly basis. The company has devised this business model on the basis of customer data and is relying on generating even more useful customer data just by the use of this flexible business model, which has potential to yield more accurate data on customers’ preferences than ‘rigid’ tariffs do. Holding on to customers is the challenge, and being able to predict churning based on machine learning algorithms an effective answer to it. Solution Approach For a detailed introduction into machine learning, the reader is referred to (Hastie et al. 2009). Machine learning, in principle, refers to the study and construction of algorithms able to learn from and make predictions on data, thus being beyond static program instructions and rather leading to data-driven predictions or decisions, based on a model built from sample inputs. No matter what procedure or application, the common element is always a decision rule that leads to separation of data into subgroups, possibly repeated hierarchically. For instance, in the iD Mobile Ireland case, data on call record information (duration, start time, number of users involved, and call type) was used to draw conclusions regarding behavioral patterns. One issue here was to decide if payment delays of some users were due to fraudulent behavior or unsuspicious reasons. A machine learning algorithm would be trained on a sufficiently large subset of the data to ‘learn’ how to separate ‘good’ from ‘bad’ customers, entirely data-driven. Another directly marketing-oriented application has been customer segmentation into a number (which number exactly is determined data-driven) of classes defined by willingness to pay—not binary, as in the fraud-issue, but into several classes in order to indicate spending habits and status and likeliness of willingness to spend more—one application of which is individually targeted marketing. (Source Dullaghan and Rozaki 2017)

6.8 Further Reading For further insights into the marketing section there are some starting leads to report: Jobber and Ellis-Chadwick (2012) are providing a general textbook on recent issues in the field and extending standard questions.

128

T. Neukirchen et al.

An application-oriented introduction to Monte-Carlos simulation is e.g. Robert and Casella (2009). Insights into the state of the art world of structural equation modelling are provided by Hair et al. (2016). A future-oriented perspective on marketing in times of social media is provided by Tuten and Solomon (2017). An interesting—historical and autobiographical—read on the subject is also Kotler (2017).

References Chapman C, Feit EMD (2014) R for marketing research and analytics. Springer, London Chapman C, Feit EMD (2015) R for marketing research and analytics. Springer, London Coyle JJ, Bardi EJ, Langley CJ (1988) The management of business logistics. St. Paul, SouthWestern Curry B, George KD (1983) Industrial concentration: a survey. J Indust Econom 31(3):203–255 Diez DM, Barr CD, Çetinkaya-Rundel M (2014) Introductory statistics with randomization and simulation. https://www.openintro.org/stat/textbook.php?stat_book=isrs Dullaghan C, Rozaki E (2017) Integration of machine learning techniques to evaluate dynamic customer segmentation analysis for mobile customers. Int J Data Min Knowled Manag Process 7(1):13–24 Farris PW, Bendle NT, Pfeifer PE, Reibstein DJ (2010) Marketing metrics: the definitive guide to measuring marketing performance. Pearson, New Jersey Gansser O (2014a) Marketingplanung als Instrument zur Krisenbewältigung. In: Krol B (Hrsg.) ifes Schriftenreihe, Band 9, Essen 2014 Gansser O (2014b) Wie Werte unsere Käufe beeinflussen. http://www.fom.de/forschung/institute/i nstitut_fuer_empirie_und_statistik.html#!tab=publikationen-2 Gansser O, Füller SR (2015) Präferenzprognosen mittels Conjoint—Analyse—eine Fallstudie mit Choice-Based-Design. In: Gansser AO, Krol B (eds) Markt- und Absatzprognosen. Springer, Wiesbaden Grue B, Ludvigsen J (2006) Decision factors underlying transport mode choice in European freight transport. In: Proceedings of the European transport conference ETC Hair JF, Hult GTM, Ringle Christian CM, Sarstedt M, Thiele KO (2016) Mirror, mirror on the wall: a comparative evaluation of composite-based structural equation modeling methods. J Acad Mark Sci 2017:1–17. https://doi.org/10.1007/s11747-017-0517-x Hastie T, Tibshirani R, Friedman J (2009) The elements of statistical learning, 2nd edn. Springer, New York Jobber D, Ellis-Chadwick F (2012) Principles and practice of marketing. McGraw-Hill Higher Education, London Jones E (2003) An application of game theory to supermarket pricing. J Food Distrib Res Food Distrib Res Soc 34(01):56–62 Kotler P (2017) Philip Kotler: some of my adventures in marketing. J Historic Res Market 9(2):203–208. https://doi.org/10.1108/JHRM-11-2016-0027 Kotler PT, Keller KL (2015) Marketing management, 15th edn. London, Pearson Malhotra NK (2009) Marketing research: an applied orientation, 6th edn. Prentice Hall, London Moutinho L, Meidan A (2003) Quantitative methods in marketing. In: Baker MJ (ed) The marketing book. Heinemann/Chartered Institute of Marketing, London Nash JF (1950) The bargaining problem. Econometrica 18(2):155–162 Robert CP, Casella G (2009) Introducing ‘Monte carlo simulation with R. Springer, London

6 Marketing Concepts and Instruments in Supply Chain Management

129

Solomon MR, Marshall GW, Stuart EW (2017) Marketing: real people, real choices, 9th edn. Prentice Hall, New Jersey Tuten TL, Solomon MR (2017) Social media marketing. Sage, London

Chapter 7

International Trade, Global Supply Chains and Compliance Albert W. Veenstra

Abstract International supply chains require the movement of goods across borders. While moving goods internationally is already operationally complex, the regulatory requirements that need to be met complicate this further. In this chapter, we explain how and why international regulatory issues enter the supply chain. At the basic level, we discuss in some detail concepts such as arm’s length trading, origin, customs declaration and the role of trade agreements. We then introduce, as part of the advanced level, the basic principles of customs supervision in supply chains. Most customs regulations contain principles that are designed to facilitate international business and allow for the postponement of duty and tax payment to the right time and place. In the supply chain literature, these concepts are virtually unknown, but they play an important role in the design of international supply chains in practice. We also introduce a new vision on customs enforcement that was developed as a spin off of several European research projects. In this vision, so-called trusted tradelanes are introduced as the next step in supply chain compliance. We end with the state of the art discussion on the design of trusted tradelanes and the necessary and sufficient conditions that need to be met to achieve this trusted tradelane status. Throughout the chapter, we have included case studies from practice to illustrate specific insights, or offer a basis for further discussion.

7.1 Introduction This chapter addresses the international dimension of doing business. We will look at the way businesses organize themselves in an international context, and we will investigate what that international dimension adds to the complexity of managing a supply chain. This way of looking at enterprises will require us to investigate the link A. W. Veenstra (B) Eindhoven University of Technology, Eindhoven, The Netherlands e-mail: [email protected]; [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_7

131

132

A. W. Veenstra

between supply chain management and international trade. A basic understanding of this link will help to understand the myriad of compliance requirements that businesses encounter when doing business in an international setting. In this chapter, we focus on compliance requirements related to the border crossing movement of goods. The government agency involved in setting and enforcing these requirements is Customs. In many countries, customs is typically part of the Tax Authority. Historically, customs supervision was largely tax related, but today it covers a wide range of other areas, such as safety and security, environmental impact, health, and food safety. In fact, our way of looking at the organization of goods flows in an international setting is fairly generic, whatever the specific compliance requirement may be. Our approach begins with the definition of a global supply chain. In this chapter, we take the “Unionist” approach of Larson and Halldorsson (2004) to supply chain management: supply chain management encompasses logistics management, but it is more than that. Supply chain management deals with the choice of suppliers, manufacturing locations, customer markets, and the required performance to satisfy customer demand and generate value for shareholders. We will explain that this view results in a hierarchy of decision making that is important to understand in the light of compliance. The remainder of this chapter is organised as follows. We first discuss our supply chain concept. We then introduce compliance as a management issue, and offer some specific insights on existing customs regimes. After that we present a new vision on customs enforcement, that explicitly recognises the so-called trusted tradelanes. We end with a discussion on the design of trusted tradelanes and a case study that offers further insight in the status of trusted tradelane.

7.2 A Definition of the Global Supply Chain The main purpose of businesses is to create value for customers. They do this by offering the best product to the customers, at a competitive price, and against low costs. In the past several decades, we have seen a development of companies crossing borders to look for customers, for markets with better prices for their product, for new, cheap and better suppliers, and for areas where cost savings in manufacturing could be realised. Many companies have expanded their presence in the world in order to create better value for their customers, but ultimately for their shareholders as well. We, and many others, have called this process globalisation.

7.2.1 Defining the Supply Chain Before we can discuss the impact of the international dimension, we first need an understanding of the interaction between supply chains, logistics and transportation.

7 International Trade, Global Supply Chains and Compliance

133

As we have said, we follow Larson and Halldorsson (2004), who characterise the so-called unionist approach that embeds logistics within supply chains. This is in line with the well known definition of supply chain management of Simchi-Levi et al. (2007): “Supply chain management is a set of approaches utilized to efficiently integrate suppliers, manufacturers, warehouses and stores, so that merchandise is produced and distributed at the right quantities, to the right locations and at the right time, in order to minimize system-wide costs while satisfying service-level requirements”. Note that this definition implies supply chain management is an integrative business function that encompasses logistics, by referring to the well known “Rs” definition of logistics (delivering the right quantities, at the right place at the right time, and so on). In other words, decision making at the supply chain level—about which suppliers to employ, which markets to target, where to manufacture products— determines which logistics problems occur and within what service level requirements these problems need to be solved. For our purposes, we wish to add a further level to this model of supply chain management and logistics, and that is transportation. In much the same way as supply chain management determines logistics, logistics management determines operations and transportation. In logistics management, the decision making is about the desired flow of goods: in what quantities to ship, how fast transportation needs to take place, where to store, and where and how to transship. Once these decisions are made, the actual transport operation must be addressed. We thus get a model with three levels, in which we depict the outcome of each decision making process as a network. This is represented in Fig. 7.1. We observe that at the supply chain level, the nodes are supplier locations, manufacturing locations, warehouse locations and market/customer locations. The links are logistics chains. At the logistics level, the nodes are factories, logistics centers such as consolidation facilities, cross docking centers and ports or airports, and the links are transport chains. At the transport level, the nodes are all the logistics facilities as well as transfer points between modes, and the links are individual transport mode connections. The hierarchy works top-down by setting requirements and narrowing the decision making space for the level below. It also works bottom-up in the sense that physical restrictions, such as the presence of rivers, ports or border crossings, or logistics facilities, need to be taken into consideration at higher levels. An illustration why this hierarchy is important for our purpose is explained next. Border crossings typically feature at the transport level. A border crossing is a point in the transport chain where governments on both sides of the border will—usually independently—require information on what is shipped, on whose behalf, and if certain formal requirements have been fulfilled. The transport operator, who may not be the owner of the goods, but is working as a contractor for either a buyer or seller or their representatives, needs to be sufficiently informed to satisfy these requirements. He or she also needs to be able to submit information at the government agency’s request in the right format, on paper, or in digital form, as the case may be. This depends strongly on the local circumstances and the type of goods moved. It is clear that there is a good chance that information asymmetries arise here. At the supply

134

A. W. Veenstra

Fig. 7.1 Supply chain-logistics-transportation management hierarchy

chain level, parties know exactly what good is produced and what it consists of, but this level of detail may not be transmitted through the hierarchy. At the transport level, parties may know exactly what the local requirements are, but they may not know how to obtain the right information from either the logistics or the supply chain level. As a result, the transport operator may not be fully informed, and will not be able to satisfy all formal demands for information. The consequence is delay at the border, subsequent delay in delivery of the goods, and thereby uncertainty is introduced in both the logistics chain and the supply chain. It is well known from basic inventory management theory (see the chapter on the bull-whip effect in this book) that this translates into higher inventories in the chain, and therefore ultimately higher costs of the goods, a phenomenon known as the bullwhip effect. See also Chap. 14, Fig. 14.7, for an illustration, and Chap. 20, Sect. 20.4.1, for remedies to overcome the bullwhip effect. We can represent the problems created by distance and the restricted information flow using Fig. 7.2. An international transaction has two sides: a buyer and a seller, or a supplier and a manufacturer, to stay in supply chain terms. These divide the responsibilities of transportation, by each appointing a local representative who will organize the local logistics chain, and by identifying an inter-continental transport operator. This transport operator often determines the route and specific ports or airports to visit. The buyer and the seller will ship goods from and to their physical locations, such as factories and warehouses. In many cases these may be owned or

7 International Trade, Global Supply Chains and Compliance

135

Fig. 7.2 A stylized international transaction

operated by third parties. This results in a stylised picture of a singe goods transport, illustrated in Fig. 7.2. Observe in Fig. 7.2 that there is quite some distance, measured in network links, between the buyer and the local operators the seller has hired as part of this transaction: two links from the buyer to the seller’s freight forwarder, three to the local transport operator, and four to the port terminal of departure. In the same way, there is a distance between the seller and the final recipient of the goods on behalf of the buyer. In logistics and supply chains, parties interact with their direct partners. Information travels poorly across multiple links. It is therefore not uncommon that a buyer or recipient does not know what factory its goods are coming from, who has stuffed a container, who has transported the goods from factory to seaport, or exactly what transport operator has to fulfill the border-crossing formalities.

7.2.2 The Mechanism of International Trade The complexity of the international movement of goods are substantial (see also Hausman et al. (2010), who document this complexity in quite some detail), and still companies have been globalising their operations at a considerable pace. While we spend much time qualifying globalisation from a societal point of view, from a generic business management point of view, this development is a very natural progression. Companies simply look for value-creation opportunities abroad. In many cases, it is a seemingly straightforward management decision to enter other countries to seek new markets, or to achieve costs savings by buying or manufacturing materials or products cheaply abroad. At the same time the world is still organised in (more or less) independent national states. Every state has its own jurisdiction, government and culture. As a result, crossing a border from one state to another results in changes in rules, regulations,

136

A. W. Veenstra

customs, and requirements. In principle, there is no unhampered flow of goods, money, people or even information across these borders. International trade has found a way of dealing with the problem of sending goods around the world a long time ago (see also the case study in Chap. 3). International trade involves two parties wishing to exchange a good, one of them is the seller, and the other the buyer. As soon as the sale is agreed, it follows that buyer and seller have to arrange the transfer of the good.1 Some of the more basic mechanisms in trade have been in existence at least since Roman times: documentation describing sender, receiver and cargo, that accompanies the cargo in transit, basic mechanisms of taxation at check points such as borders and so on. Many of these mechanisms are still in place, but have been codified in international conventions, such as the United National Convention on Contracts for the International Sale of Goods, and related conventions for the international transportation of these goods, for instance the Hague, Hague-Visby and Hamburg Rules for ocean transport, and the CMR2 convention for international road transport. The core concept in these conventions is the description of the basic transport documentation, the information it should contain, and the rights and responsibilities of the various parties involved in contracts. Another important mechanism that has developed over time is the way trading parties divide the responsibilities for transportation through a set of so-called Incoterms.3 Nowadays, this is a set of standard arrangements governed by the International Chamber of Commerce (ICC). These arrangements stretch from the seller taking all responsibilities for transport and formalities, to the buyer taking all these responsibilities. There are eleven different stages, that are marked by a specific point in the international transport chain where responsibilities for arranging transport, insurance and ownership of the goods can be transferred between buyer and seller. These are illustrated in Fig. 7.3. In general, in international trade, the Incoterms Free on Board (FOB) and Cost, Insurance, Freight (CIF) were used predominantly. These two maintain the local transport responsibilities with the buyer and seller, and allocate the international transport link to either the buyer (FOB) or the seller (CIF). For the transportation of containers, however, these incoterms are not very suitable, and incoterms such as CFR, CPT and FCA are more commonly used. In circumstances where the sales contract occurs between parties that know each other, or are part of the same conglomerate, the more extreme Incoterms will generally be used: Ex Works (EXW), where the receiving party takes care of all transport related activities, or Delivery, Duty Paid (DDP) where the sender is fully responsible. In these cases, one party will control the entire transport chain, and this can be advantageous for a number 1 There

is an important exception to this: the pro-active movement of spare parts, as a result of warranty agreements. In this case, there is no contract of sale with a buyer and seller that drives the international movement of goods. Many elements of these transactions remain similar to a sale and purchase arrangement, however. Another example is stock movements by e-commerce companies. 2 CMR stands for Convention Relative au Contrat de Transport International de Marchandises par Route. 3 Incoterms is an abbreviation of International Commercial Terms, referring to a set of terms of sale that are accepted worldwide.

Fig. 7.3 Incoterms 2010 (International Chamber of Commerce ICC)

7 International Trade, Global Supply Chains and Compliance 137

138

A. W. Veenstra

of reasons: the entire service can be purchased as a package from a single service provider, security arrangements can be made in the entire chain without handovers in the chain of custody of the goods, and so on. These two incoterms should, however, only be used if buyer and seller are licensed to carry out customs declarations in each other’s country. The importance of Incoterms is that they link the commercial transaction to a transport arrangement. Given a specific choice of incoterms, both buyer and seller know what to do to finalise the transaction and exchange the goods for payment. They will each make the proper arrangements with service providers, and as a result, goods go from seller to buyer, and money flows seamlessly from buyer to seller. Given that buyer and seller want to make this exchange work, but may not trust or know each other, transportation has been given a further role in this exchange. To avoid the deadlock that the buyer will demand payment first before sending the goods, and the seller demanding the goods before sending payment, transportation plays a role in sending the goods and the payment “at the same time”. The handover of the goods by the seller to a third party (the transport operator) is formalised in the international trade process to be used as proof that the goods are out of the hands of the seller. As a result of this, the buyer will also relinquish funds for payment to the seller’s bank. Only when the goods are received by the buyer, will the payment be made to the seller. This process is done by so-called letters of credit, and this is still a mechanisms for certain international trade transactions. Case Study: Grain Trader Grain Trader BV is a trader of starches and grains. Its business model is to buy product from growers and traders around the world, and then sell these products somewhere else in the world. Part of the transaction is usually the intercontinental transportation of the product. Grain Trader buys its products under FOB conditions and sells under CIF conditions. This means that Grain Trader becomes responsible for the ocean transport part of the transaction against both buyer and seller. In some cases, however, there are profit-making opportunities in local transportation. This might be the case when Grain Trader buys or sells large volumes to or from several different parties. In that case, it might have purchasing power for a local transport solution that is better than the local parties could offer individually. Grain Trader then adapts the incoterms of the purchasing or sales transactions to be in a position to arrange local transport as well. Consider a case of buying grain in the United States of America which is a major supply country for Grain Trader. This grain will be sold in Germany, but for a relatively low price. Contrary to the standard approach, in the USA, Grain Trader has a good deal on barge transport from the grain growing areas to the port of New Orleans. The customer is willing to pay an additional fee for this transport leg, as well as for the delivery of the grain to its storage facility in Germany. For local European transport, Grain Trader also has access to

7 International Trade, Global Supply Chains and Compliance

139

relatively inexpensive transport options, such as rail and barge, all of which can travel directly into the customer’s premises. As a result of this, Grain Trader will buy the grain ExWorks and sell to the customer DeliveryAtPlace, where the named place is the buyer’s premises. By charging the buyer for transport on both sides of the chain, and since these transport options are attractively priced for Grain Trader, it is able to compensate the poor trading profit on the grain itself.

7.2.3 The Global Supply Chain We have described the expansion of business to other countries and other parts of the world as a natural process for many companies. The field of International Business Management offers a growth model of the way businesses will enter international markets. This is usually a four-step model: (1) regular export, (2) appointing a sales agent abroad, (3) starting a joint venture with a foreign party, and (4) developing a wholly-owned subsidiary. This type of models represents the development of internationalising sales. There are also other business activities that can be internationalised, such as purchasing, manufacturing, or administration. These will also follow similar patterns of starting small, without a foreign presence, and developing into a more balanced multi-country enterprise. We have already remarked on the differences in jurisdiction that enterprises encounter when they begin to operate in other countries. These jurisdictions are still confined to the borders of the respective countries. There are also supranational rules that apply when companies are combining activities in different countries into the manufacturing of their final product. One such rule has to do with the transaction to send goods between two parties that belong to the same international enterprise. This rule is called “arm’s length trading” and refers to the need to set a fair market price for any good that is moved internationally, between whatever parties.4 This rule avoids the practice of companies to manipulate prices of components, parts, and final products in subsidiaries in order to concentrate profits or losses in specific countries. While this rule applies to physical goods, there are many other ways in which companies can still redistribute profits in their international ownership structure, to the displeasure of politicians who see large conglomerates such as Apple and Starbucks paying very little tax anywhere. The arm’s length trading principle effectively makes every transaction in a supply chain a real trade transaction, which functions in much the same way as the general international trade transaction we described above. This suits many international companies quite well, because it helps to distribute responsibilities fairly for sales, 4 The supranational basis for the arm’s length trading principle is complicated, but its formal codifi-

cation might be traced back to the General Agreement of Trade and Tariffs (GATT) of 1994, article VII, which is now included in the WTO Valuation Agreement. It is also adopted in the OECD Model Tax Convention. Shifting profits is still a contentious topic, however, as a result of many bilateral tax agreements that allow corporations to distribute profits to low tax jurisdictions through other means, such as royalties and brand management fees.

140

A. W. Veenstra

profit, and costs among business units or profit centers. There are some simplifications because parties in the same conglomerate know and trust each other. Letters of credit are not often used in inter-company transactions and the intercontinental transport transaction can also be simplified because it is part of a global transaction between the company and transportation service providers. But the principle remains the same. A global supply chain may thus be characterised by suppliers being in one country, manufacturing in another, and logistics facilities and the final customers again in some other countries. A company operating such a supply chain runs the risk of paying taxes and duties on all border crossing goods flows. Because some of these flows are for materials and components that will be incorporated into products, these payments may compound into a serious cost component, if nothing is done. Value-added tax (VAT), for instance that has to be paid for goods entering any country, is usually somewhere between 15 and 25%. If a product consists of a basic material that is moved across a border, made into a component, crosses another border, assembled into a final product, which is moved across a border into a regional distribution center, and then moved across a border to a customer, that VAT percentage is applied to the consignment four times. VAT will thus become a significant part of the cost structure of the final good. Luckily, there are solutions for this problem. The general principle for all these solutions is that tax and duty payment can be returned, or better, postponed until the “right” moment. This can be the point in time when final delivery to a customer takes place, or specific points in time when clear value is added. We will discuss these solutions in the next section. We dedicate the remainder of this section to another theme that plays a role in the discussion on global supply chains, and that is the tendency of countries to align their legal frameworks for trade, thus removing potential barriers for trade that arise from crossing borders. This development results in so-called free trade agreements (FTA) that offer preferential treatment for trade between parties in the agreement, reducing taxes and duties, and possibly removing other supervision-related handling of goods at borders. Just like trade, trade agreements have been around for centuries. For an extensive discussion on the development of trade agreements and their relation to customs unions, see Viner (2014). Trade agreements follow from the so-called most favoured nation (MFN) principle. This is a rule that states that countries who accord this principle to another country must bestow upon that country the same level of treatment as they offer to their most favoured nations. All members of the World Trade Organization offer MFN status to each other. The only exceptions are preferential treatment of developing nations, regional free trade agreements and customs unions. In these cases, more extensive “favouring” of other nations can take place, up to the level of charging zero duty rates, and removing documentary requirements. The European Union has been a customs union since 1993, which means that there is a common customs policy, and free movement of goods between all Member States. NAFTA, between the United States of America, Canada and Mexico is a regional trade agreement. These three countries do not have a common trade policy or a common customs policy, and are therefore not a customs union. The growth of these trade agreements is illustrated in Fig. 7.4.

7 International Trade, Global Supply Chains and Compliance

141

Fig. 7.4 Overview of trade agreements over time (WTO Secretariat). The legend refers to trade agreements between two types of countries: developed and/or developing. PTA stands for Preferential Trade Agreement

The major benefits of trade agreements are that they result in lower costs when crossing borders and in less uncertainty, because of the simplification of procedures and requirements. In the context of supply chain design, it thus makes sense to take the existence of trade agreements into consideration when making decisions at the supply chain and logistics levels in our hierarchy. Fitting a supply chain operation to trade agreements is the subject of a case study on the manufacturing of the Renault Logan car, see Lee and Silverman (2008). The significant rise of the number of trade agreements, as depicted in Fig. 7.4 means that this process of taking advantage of trade agreements in the construction of supply chains is also becoming more and more complex. The consequence is that many companies do not take full advantage of trade agreements, and are therefore paying too much taxes and duties. An important element of the enforcement of trade agreements is the formal proof that the goods are coming from a specific country that is part of the trade agreement, or of some other preferential arrangement. This proof is often a so-called certificate of origin (CoO). This is simply a document that is provided by a country’s government or some representative agency (this is often a Chamber of Commerce), that is accepted as proof by customs in another country. This document has to be presented at the border, and has to be attached to other formal declarations related to the goods. In Europe, for countries with which Europe has a trade agreement, the so-called EUR.1document exists for this purpose. Europe requires CoOs from many countries for the import of goods from these countries.

142

A. W. Veenstra

Case Study: Logistics and Origin A jeans manufacturer in Europe produces jeans and related products in factories in China, Bangladesh and the Philippines. For Bangladesh and the Philippines, Europe has preferential treatment arrangements in place that require a certificate of origin with every shipments. The shipments arrive in the Netherlands, where they are unpacked, and consolidated and repacked at shop level for further distribution. To save costs on the expensive packing and repacking operation in the Netherlands, the jeans manufacturer has decided to develop a cross-docking facility halfway between Asia and Europe, on the Indian subcontinent. This lies along the shipping route, and offers less expensive labour, within the confines of a free zone. Goods entering and leaving the free zone are exempted from taxes and duties. The consequence of this cross docking operation is that the origin certificates from Bangladesh and Vietnam no longer match the consignments that enter into Europe, because the original consignments are now distributed across shipments that are ordered at the destination country and at shop level. As such, customs in the Netherlands cannot determine whether the goods are covered by an origin certificate, and therefore preferential import duties do or do not apply. This may result in duties being charged of up to 12% on the value of the products, instead of zero duties. There are currently two possible solutions for this problem. One is to obtain origin certificates for each individual product, or at least for the set of products that are destined for the same store. This is a cumbersome solution, because countries usually charge for each CoO, and many countries do not allow many different CoOs to be produced for (parts of) a single international shipment. At the very least, this will add extra cost to the product. The second option is to develop an administrative reconciliation of origin certificates and shipments with customs in the Netherlands. While this might be possible, it puts extra pressure on the operation in the cross docking facility to avoid leaving products behind that may obscure the reconciliation. Also, customs is not very confident that the original link between products and certificates can be verified. The European Union offers a third solution, which is the registered Exporter System (REX) that came into force in 2016. This allows exporting countries to the EU to register their companies as approved exporters. These companies can then issue CoOs themselves. This would facilitate the first solution to work in practice, without driving up the costs. Both the Philippines and Bangladesh have opted to introduce REX in 2019.

7 International Trade, Global Supply Chains and Compliance

143

7.3 Compliance Requirements in International Business The word “to comply”, according to the Oxford English Dictionary, means “to act in accordance with a wish or command”. This explanation already encompasses two types of compliance: one that results from a wish, and one that results from a command. In other words, compliance can be voluntary or compulsory. Voluntary compliance is often introduced via corporate social responsibility schemes, where businesses promise to ban child labour, products for which animals need to be killed unnecessarily, or fake materials, or to avoid unnecessary waste and to pay fair wages in their entire enterprise. These are voluntary in the sense that the company can change these rules. While the rules are in place, however, they may be felt to be as compulsory as government regulations to many in the company. Compulsory compliance typically arises from regulatory or other formal requirements that are set by governments. From these two options, a third one may follow: companies imposing rules on other companies. A company that promises to avoid waste may impose this same condition on its suppliers, service providers and sometimes even customers. Depending on the purchasing power of this company, the parties at the receiving end may consider these demands to be almost equal to government regulations. For the voluntary schemes, or schemes introduced by societal interest parties such as Greenpeace and other non-governmental organizations (NGOs), labels are often introduced. Greenpeace, for instance, monitors energy use of information technology (IT) companies in their Clicking Clean report. For many agricultural products, there are sustainability and good governance labels, such as UTZ Certified,5 or Max Havelaar6 in the Netherlands. Another well known example of sustainability ranking is the Dow Jones Sustainability™Index (www.djindexes.com/sustainability), which some companies refer to in their corporate sustainability reports. In the compulsory schemes, the main purpose is to offer insight on the flow of goods to government parties in order to enable them to perform their formal inspection duties. In this chapter we will largely focus on compulsory compliance, although we will see that the line between compulsory and voluntary can become blurred.

7.3.1 The Introduction of Global Security Schemes For security in supply chains, the terrorist attacks in the US on September 11, 2001 (commonly known as “9/11”) provided a major boost in the introduction of security schemes. At the same time, supply chain security was not a new topic for many companies. One case in point is the supply chain disruptions at Ericsson and Nokia in early 2001, which became famous because of the strikingly different impact of the disruption on the two companies. See for an exposé, Norrman and Jansson (2004). 5 UTZ

comes from the Guatamalan Mayan word utz kapeh, which means good coffee.

6 The Dutch foundation Max Havelaar aims to get better prices for commodities, such as coffee and

cocoa, for small holder farmers. These commodities are labeled with the Fairtrade logo.

144

A. W. Veenstra

Another security scheme that already existed before 9/11 is the Technology Asset Protection Association (TAPA), the membership of which consists mainly of technology companies that are confronted with a high level of crime, such as theft from warehouses, hijacking of trucks, and so on. This organization already had an audit scheme for warehouses and trucking operations since its inception in 1997. Furthermore, standard supply chain management literature on inventory management or production planning has long contained notions of running out of stock, and the possible disruptions in supply chains of machine breakdowns. Finally, the role of customs agencies around the world has been to inspect cargo in order to avoid counterfeiting, allowing dangerous medicine, or unsafe children’s toys to enter the country. The events of 9/11, however, added an extra dimension to these discussions and insights: supply chains could become disrupted or be misused as a result of breaches in security. Threats such as carrying (parts of) a dirty bomb in a container or using a ship to disrupt a port were formulated, and the identification of these potential threats led to the development of a large portfolio of security measures. We will briefly discuss the most important ones here, but see for a more elaborate discussion, for instance Giermanski (2012) and Chap. 25 of this volume. The first measures that were introduced unilaterally by the United States of America was the so-called Container Security Initiative (CSI). This is a set of rules that combined the principle of “pushing out the border” and the introduction of technological solutions to increase the visibility of container traffic. The pushing out the border principle is the notion that the importing country’s customs organisation should be informed earlier about what will be transported to that country, preferably before departure of the plane or ship from the country of origin, and, furthermore, should be in a position to deny shipment of these goods. Part of this was the placing of US customs officers in ports overseas to coordinate and support this process. For ocean shipping, this part of the CSI was called the 24-hour rule. The notion “24hours” refers to the time frame of 24-hours before departure of the ship in which US Customs needed to be informed, and would evaluate the cargo to be loaded onto the ship. This could lead to a no-load notification, in which case the cargo could not be loaded, or if it was already loaded would definitely be off-loaded and detained in the port of destination. In air freight, a similar arrangement was introduced, but with a shorter time period, due to the faster pace of air transport. A second major initiative was the International Ship and Port Facility Security Code (ISPS), which became a separate chapter in the UN Safety of Life At Sea (SOLAS) convention. This code introduced global security levels for ships and port terminals, and a mechanism that ships could transfer higher security levels to their ports of call. The code further contained provisions on how to deal with increased security levels, through a security plan, and a designated security officer. A final component in the security schemes being introduced after 9/11 was the Customs-Trade Partnership Against Terrorism (C-TPAT). This was a scheme directed at shippers, who were supposed to develop an internal security plan to counter the interference of terrorist in their supply chain, and at the same time, impose these same rules on their supply chain partners. C-TPAT thus worked as a security proliferation mechanism throughout the world, touching all companies having relationships with

7 International Trade, Global Supply Chains and Compliance

145

companies based in the US. A drawback of the C-TPAT scheme was that it only applied for cargo flows to the USA, because these fell under the jurisdiction of the department of Homeland Security in the US. Export flows fell under the guidance of the US Department of Commerce, which had a different view on the importance of security. This difference hampered discussions on reciprocity with, for instance, the European scheme of the Authorized Economic Operator, which covers both import and export. Many other countries developed security schemes as a response to these American initiatives, to protect their trade relationships with the US. Some of these, such as the South American Business Alliance for Secure Commerce (BASC), already existed, but gained new importance. In Europe, the developments and initiatives from the US side were closely watched, and many ports in Europe were early adopters of the CSI scheme and ISPS rules. The European Union (EU) followed somewhat later with its own security arrangement, which was called the Authorized Economic Operator (AEO). This certification scheme was introduced in 2007, after years of discussions and internal negotiations, and served essentially two purposes: it is the formal registration of companies in Europe that import or export goods, and that wish to use certain simplifications in customs procedures. At the same time, it is the security certification scheme for European companies active in international trade. Companies can choose to certify for either the one purpose, or the other, or both. The application procedure includes a self-assessment by the company, and an audit by the customs agency in their member state. Every 3 years, the audits are renewed and the certificates re-affirmed. As a complement to CSI, the European Union has also introduced its own variant, the so-called Entry Summary Declaration (ENS). This is a declaration about the goods on ships and airplanes that need to be lodged 24 hours before departure of the ship and 4 h before the departure of the plane to the first port of call in the European Union. Just as with CSI, no-load decisions may be issued for cargo that may not enter the European Union. When we review all these security schemes, we can observe that the distinction between voluntary and compulsory schemes becomes blurred. CSI, for instance, is a scheme directed at ports and port authorities. As such, it was voluntary, and many ports in China maintained traffic to the US while not being a CSI partner. The Port of Shanghai, for instance, held out until 2005 to become a CSI partner, as one of the last major ports in the world. However, as soon as the Port had decided to become a partner, satisfying the 24-hour rule became an obligation for all port users with traffic to the US. In the same way, C-TPAT is a voluntary scheme, but the obligation to impose the same rules as the C-TPAT member on all supply chain partners introduced an obligation for other companies. ISPS, on the other hand became part of a ratified UN convention, and as such, in many countries was a regulatory obligation. AEO in Europe is still a voluntary scheme, although many companies cannot afford to opt out of being an AEO. In commercial contracts, being an AEO is often a qualifying criterion for companies. ENS is part of the European customs regulation, and as such has a more formal status than CSI. It is a regulatory requirement that is imposed on the transport operators, just as ISPS.

146

A. W. Veenstra

With many of these certification schemes, the development is similar to ISO9001: there is a certain first mover advantage, when the certification is new, and it is a means to obtain a distinction from competitors. After a while, it becomes a market standard, and the real distinction disappears. There is even a threat that companies that do not take the certificate seriously, may fail the periodic review. There is also some evidence that adopting certification, such as ISO9001, has more general benefits in terms of improving the general operations of a company, see Anderson et al. (1999). At the same time, some companies may want to develop a new or enhanced version of a certification scheme, to better match their higher levels of security and control compared to the market standard. This is currently the case with the European AEO certificate, which reflects a fairly average security standard that is not a realistic description of the security standard of, for instance, high tech companies.

7.3.2 Customs Regulation Customs regulation is in its core a compulsory framework. In Europe, there is a common customs policy, that applies equally to all companies in all member states. At the same time, there is a high degree of modularity and company-specific fine-tuning in any customs regulations framework, due to the possibility to apply for so-called simplifications. These are modified, simplified rules and obligations for companies that can show a higher degree of “control”. Companies that are in control perform more checks and inspections themselves, audit their partners, gather and store more data, and, importantly, can communicate about what they do to customs agencies. Simplifications are awarded through additional licenses that need to be maintained and approved periodically. These licenses come with other, more administrative obligations, and result in less physical interference of customs in the flow of goods. The general principle of any customs framework is simple: when goods cross borders, duties and taxes are due, and customs will verify the goods against a list of requirements related to health, safety, environment, and economic protection. It is possible, however, to postpone these obligations to the proper time and place, in order to avoid paying duties and taxes twice. The purpose of these simplifications is to essentially tune customs regulations to the supply chain of companies. Companies may manufacture products by assembling many components from different areas, and then proceed to sell these products abroad. We have already noticed the threat to compound duties and taxes in such a supply chain. In order to avoid this, customs regulations contain the possibility to postpone payment of duties on imports if there is an expectation of re-export. In fact, there are a number of these “simplifications” in most customs regimes around the world: • Transit • Bonded Warehouse • Free Zone

7 International Trade, Global Supply Chains and Compliance

• • • •

147

Inward Processing Relief Processing Under Customs Control Outward Processing Relief Temporary Import/Export

Before we explain each of these regimes, we should first introduce the importance of certain physical areas under customs regulation. In principle, the role of customs agencies is to stop, inspect and charge goods at the border, before they can enter the territory of a country. In ports and airports, this is almost physically impossible. Therefore, there is a provision in customs regulations that there can be securely guarded areas in a country where transport operators can unload cargo. The cargo cannot go anywhere without further notification to customs, but at least it is allowed to “land” in the country’s territory. This type of area is called an Temporary Storage Area (TSA) or facility. Port terminals, or airside freight handling stations can apply to be such a TSA. If goods are brought into a customs area, all checks are performed by customs and if all duties and taxes are paid, then the goods are free to move around in that area. This is called Free Circulation. Transit is a customs arrangement under which goods can be transported across a territory while duty payment, and often, customs control, has not taken place. This procedure is used for transportation between a TSA and another customs approved area, such as a bonded warehouse. It is important for customs to be able to track the goods from their secure origin to the secure destination to ensure goods which duties have not been paid, cannot “disappear”. For this purpose, the regulation often includes maximum transit times for specific means of transport, and the European Union maintains an information system in which all transit requests are logged and cleared upon arrival. A bonded warehouse is a warehouse in which goods maybe stored under suspension of duty. This might be advantageous because the company does not know if the goods will remain in the customs area, or have to be exported again, for instance to countries just outside the customs area. In Europe, these countries might be Russia, Turkey, Norway, and Switzerland, which are not part of the Union. In addition, the company may want to bridge the time between arrival of the goods and final sale. The immediate payment of duties and taxes is a cash outflow that the company can only afford if receipt of the selling price is imminent. Many customs regimes also have the possibility of a Free Zone. This is an area that can be designated as customs controlled. Access to the free zone is then supervised by customs, and within the free zone, duties and taxes may be suspended or exempted. In many countries, other rules and regulations may be suspended in free zones. Some countries in the world, for instance, have strict ownership rules for companies: in certain industries, the majority ownership should be by the country’s nationals. In other cases, convertibility of the currency is less restricted. In free zones, countries can relax these rules without abandoning them in the country as a whole. As such, free zones are also a development mechanism for many countries. In European customs regulation, the free zone does not exist anymore. The Airport Schiphol

148

A. W. Veenstra

in the Netherlands, and the Port of Hamburg in Germany were free zones, but had to relinquish this regime as a result of the introduction of the new Union Customs Code. The solution was to expand the TSA license to many individual terminals and handling companies. Inward processing relief offers the possibility to suspend duties and import taxes, on materials and components that are going to be used in a manufacturing or transformation process, and will be re-exported. If everything that was imported is exported, no duties are payable. There is however, a risk that some of the products will be brought in free circulation anyway. In that case, duties have to be paid on the imported components. For this risk, the company should hold a bond that covers this risk. For manufacturing, this procedure is fairly straightforward. It also applies, however, to other processing operations, such as refurbishment, cleaning, repair, and so on. The exact determination of which activity qualifies for inward processing relief can sometimes be difficult. Also, the procedure is complicated due to certain materials or goods that may be consumed in the processing procedure. A catalyst is a case in point. For these goods, export cannot be proven, but the goods are also not brought in free circulation. Such complicated situations often lead to long discussions with customs agencies, and the more complicated the procedure, the more risk there is for different interpretations among customs agencies or even customs officers. There is another variant of the inward processing relief procedure, and that is processing under customs control, or inward processing under the drawback system, as it is currently called in Europe. Under this regime, goods are brought into the customs area, and will be assembled into a new final product. This regime is used for situations where the materials or components are charged a higher tariff than the final product. In this case, if these final goods are brought into free circulation, the lower tariff applies, and the difference can be claimed back. Outward processing relief is a similar construct, but then applies to goods leaving the customs area, to be processed somewhere else, and then come back. It prevents the full payment of duties for the goods entering the customs area, if it can be proven that they left that same area some time before. This mechanism is used extensively for contract manufacturing, tolling arrangements and repair or refurbishment, where parties outside the customs area are contracted to perform certain activities and goods are provided to them. These goods can be either materials, components, or a broken component that can still be repaired. This procedure can clash with certain origin related preferences, that might reduce tariffs for certain goods to zero. In that case, the complicated registration of exports that are to be re-imported may be foregone. Finally, we have temporary export or import. These are procedures meant for goods that will temporarily leave or enter the customs area and that remain unaltered. This is the case for samples, conference materials, theater stage equipment, tools for specific repair and construction work, and so on. Ocean containers and container tracking devices may also be brought into a country, laden or empty, under a temporary import license. If the goods are finally brought into free circulation, the amount of duty payment needs to be paid. For this, three elements are essential:

7 International Trade, Global Supply Chains and Compliance

149

• Origin, • Value, • Tariff. Origin is important to see if general preferences or specific country-related rules apply. These rules can be either beneficial (lower tariff) or detrimental (maximum amounts of goods that can be imported in a specific period, i.e., quota). In addition, the origin also determines if certain bi- or multi-lateral trade agreements are applicable. The value is the basis for charging duties and taxes. As such, there are specific rules to determine the value of the goods. In many supply chains, goods are traded several times: once in the purchase of the trader or manufacturer, and once in the sale to a customer. In many supply chains, there can be trades in between. This is common in commodities, but also in textiles, apparel, and other goods. To determine the value of a good is therefore not always an easy task, because it depends on the stage the supply chain is in. In Europe, the common policy is to use the value of the last sale before entering the EU. The exact application of this policy is, as yet, unclear. A second important aspect related to value is the international regime on transfer pricing. It was already mentioned before that trades, even between subsidiaries of the same company, need to take place at market prices. If goods are sent across borders without a real trade transaction (for instance, because the goods are sent from one office in the same company to the other, or because movement is related to a warranty arrangement), still a “real” value has to be determined. Based on the origin and the value, and the applicable tariff book, the tariff for duties and taxes is finally determined. In many countries around the world, the determination of the tariff on a transaction basis can be avoided by applying for a binding tariff decision of the customs agency. This is relevant for repeated imports of the same good, and for new imports of goods that have not yet been imported into a customs area. An example of the latter case is a children’s toy that can play music, record videos, make photographs, and play games. Each of these functions exists in separate toys, and they all have different customs duty tariffs. Importing this product without any previous discussion with customs will therefore result in a high degree of uncertainty as to the tariff that customs will choose to apply. Obtaining a binding tariff decision beforehand is therefore a sensible strategy.

7.4 New Enforcement Vision: Towards Trusted Tradelanes In the previous sections we described the general customs principles and regimes, that are valid around the world at the basic and advanced level. Each country may have different local rules, and in some countries, some of the regimes may not exist, but in principle, these regimes are rather generic. It remains true, however, that many countries attempt to use their particular customs policy as an economic policy

150

A. W. Veenstra

to attract foreign investment and foreign business activities. Free zones in many developing and newly industrialised countries are a good example of this. In Europe, competition between customs regimes is not easy, since there is a common European Customs policy. There are still ways to distinguish oneself from other countries within the European Union. One of these ways is the application process of customs licenses. The Netherlands was one of the first countries in Europe to award bonded warehouse licenses to logistics service providers. In many other EU Member States, this is considered impossible, because a logistics service provider is not the owner of the goods, and is therefore considered unable to keep a good account of the goods stored in the bonded warehouse. Partly as a result of this policy, the Netherlands hosts many European distribution centers managed by logistics service providers. Another way of distinguishing oneself as a country is by thought-leadership. This section will discuss the Dutch vision on enforcement that was launched in 2014, and that describes a new development path for customs agencies for the coming years.

7.4.1 A New Enforcement Vision The enforcement vision presented here contains a number of standard features of any customs enforcement approach. The vision is depicted in Fig. 7.5. There is a basic distinction between the use of technology, as represented by the technology roadmap at the bottom of Fig. 7.5, and a risk based analysis represented by the upper section of the figure. This risk based approach consists of the collection of data from various sources (represented by the funnel on the lefthand side of Fig. 7.5), and the resulting enforcement mix, represented by the pie charts, with diagrams representing different enforcement activities such as physical intervention, verification of declarations, and post clearance audits. The European customs policy has introduced a split between so-called trusted traders and unknown traders. The first, identified through the AEO-certificate, are treated differently in terms of the enforcement activities than the latter. While unknown traders can count on a relatively high percentage of physical inspections, trusted traders will more often see transaction-based post-clearance audits and system-based post-clearance audits. The technology roadmap augments this risk-based treatment of traders with an approach that aims to touch all goods flows: scanners, nuclear detection, and sniffer technology. All boxes, containers, pallets on all modes are led through these technologies. The risk based approach together with this 100% detection offers an effective, but also affordable, customs enforcement approach.

7 International Trade, Global Supply Chains and Compliance

151

Fig. 7.5 Dutch enforcement vision (Dutch Customs)

The new addition to this enforcement vision is a third category of traders. These are traders who operate in chains that completely consist of trusted traders and are characterised by information sharing, anomaly detection, and self-correction mechanisms. Such trade lanes can be awarded with an even less intrusive regime, that mainly relies on customs obtaining all the data they need from the chain information system that supports all actors in a secure trade lane. The precise conditions for a trusted tradelane are not yet fully developed. Some elements have been identified in recent research projects, however. These will be discussed in the next section.

7.4.2 Fulfilling Compliance Requirements for Trusted Tradelanes So far, we have discussed the way compliance requirements in international supply chains arise, and what customs regimes exist to facilitate the design of international supply chains. This section will address the advanced solutions that businesses have developed to fulfill compliance obligations.

152

A. W. Veenstra

Before we discuss some of these new solutions, we first have to describe the standard way in which government agencies are informed about the international flow of goods. This is done through a so-called declaration. This is a standard document that needs to be filled in by the responsible party, and sent to customs or any other government agency that has formal responsibilities in the supervision of border crossing flows. The format and content of declarations is formalised in customs law. The obligation to submit declarations is part of public law, as opposed to civil or criminal law, which means that after submission of a declaration, the government agencies can ask for as much additional information as they like, and businesses are bound to offer this information, even if this helps to confirm their guilt in submitting a wrong declaration. The declaration, in principle, is transaction based. Every shipment requires a separate declaration that needs to be lodged at the moment the transaction crosses a border. Government agencies can offer simplifications in the sense that businesses can apply for periodic declarations for certain procedures, or for declarations after the fact. Many companies, especially those with large export or import volumes, use such simplified declaration procedures. The new Union Customs Code also allows the possibility to lodge an incomplete declaration. This is practical in the case of e-commerce shipments, which are often small packages that fall below tax and duty thresholds. In such cases, a tariff code may not be required and could even be unknown to the party responsible for sending the declaration to customs. This compulsory field in the declaration can now be left blank. In the previous section, we have introduced the concept of a trusted tradelane, which is a compliance regime that covers an entire supply chain. Such a regime contains a number of components, that we will discuss in some detail below. They are: (1) the structured identification of risk in the supply chain, and (2) the transmission of data along the supply chain through the data pipeline. We will then end this section with a roadmap to develop trusted tradelanes.

7.4.2.1

Risk Identification in Supply Chains

For customs compliance, the definition of risk is crucial. In this section, we will elaborate on the perceptions of risk from a business and a government perspective. First, however, we briefly introduce risk in a general sense. BusinessDictionary.com defines risk as “a probability of damage, injury, liability or loss that is caused by external or internal vulnerabilities, and that may be avoided through preemptive action.” Risks for business In the logistics industry the risk definition follows the well known 7Rs definition of logistics itself. Risks therefore are: • Not delivering the right product, • Not delivering to the right customs,

7 International Trade, Global Supply Chains and Compliance

• • • • •

153

Not delivering the right quantity, Not in the right condition, Not at the right place, Not at the right time, and Not at the right costs.

Risks all have more or less the same structure: there is a performance variable (time, place, quality), that needs to be close to some predetermined performance target. The comparison of these two variables leads to the observation of risk (signal), and possible corrective action. In some situations, the performance variable needs to be exactly the performance target. In other cases, the performance variable needs to be at least at good as the performance target. This holds especially for time-related risks, where actions need to occur before or as close as possible to a certain due-date. After 9/11, a considerable body of literature emerged on risk identification and management in supply chains. See, for instance, Chopra and Sodhi (2004) or Jüttner et al. (2003). The purpose of much of this research was to identify, i.e., enumerate, all possible risks and to specify so-called risk frameworks that are constructs in which some sort of overview of all relevant risks could be displayed. Case Study: Container Logistics Chain Risk Identification Via the EU-FP7 project INTEGRITY (http://www.integrity-supplychain.eu), a risk identification protocol was made for a regular container logistics chain that connects a factory in China through a port in China and a port in Europe, with a final destination in Europe. Logistics service providers were asked to identify major risks in this chain. These risks turned out to match important milestones in the chain: • • • • • • • • • • •

container closure at factory, container departure from factory, container arrival in terminal, ship arrival at port of origin, container loaded on ship, ship departure, ship arrival at port of destination, container unloaded, ship departure at port of destination, container departure from terminal at destination, container arrival in warehouse.

An important milestone was “container arrival in terminal”. If this milestone was not met before the “ship arrival” milestone, the container would probably not be loaded, and serious delay at the destination would be the result. In the entire chain, the container would be tied to a specific ship. The reason for this is that the shipping lines which own the containers only release them for loading at factories when there is a booking of the container on a ship. As a result of

154

A. W. Veenstra

this, getting the container on the right ship is an important task for the local logistics service provider. From a tracking and tracing point of view, data on the milestones could be provided by service providers, such as the terminals and the shipping company. Only the data on the first two milestones remained difficult to obtain. The solution for this in the Integrity project was to use so-called container security devices, which would be GPS-enabled and could be fitted on a container door. Another major risk identified by logistics service providers was a gap between the actual content of the container and the documentation. In the retail business, for instance, this might result in promotional goods not being received, and therefore not being available in stores. Several solutions were eventually introduced for this: a penalty system for shippers who were sloppy with their container manifests, a tally man observing the loading process independently, and introducing a consolidation center with compulsory packing by the customer’s freight forwarder at origin. Risks for customs The main source for the customs risk management approach is the World Customs Organization (WCO) Risk Management Compendium (see http://www.cassandraproject.eu). This compendium defines risk as “the effect of uncertainty on objectives”. The WCO sees risk management as one component in a more general compliance management approach, that also contains the legislative framework, administrative arrangements and threat identification technology. The customs view on risk management is thus a much more encompassing approach that focuses more on the risk management approach of businesses, than on the actual business risks identified. In fact, many business risks identified, such as the container missing the ship in the port of origin, do not interest customs much. What customs is interested in is how thorough companies identify and qualify their risks, how complete they are in specifying risk mitigating measures, and where they stand on the risk management maturity scale. Looking at the case study in the previous section, customs was interested in the proof of non-tampering during transportation that the container security device provided, much more than the exact identification of the container closure time. Even with this proof, however, the crucial process of container stuffing, where contraband could be packed with the regular goods, could still not be controlled. Customs was therefore also very interested in the remedial measures that were introduced in the chain, such as the penalty system, the tallyman and the compulsory consolidation. The conclusion from this discussion is that the alignment of the perception of risk between business and government is not obvious. For operational risks, which are the bread and butter for logistics service providers and transport operators, the alignment is unclear. If we move, however, to risks related to the cargo itself, the alignment is much greater, even though this is for different reasons. Business is still interested in delivering the right goods, while customs is interested in avoiding transportation of banned goods. The verification for these two types of risks is the same, however: does the shipment contain what the documents state?

7 International Trade, Global Supply Chains and Compliance

7.4.2.2

155

The Data Pipeline

The previous section identified the need, in logistics or supply chains, to regularly verify actual information against expected values. This information can be operational milestones, or information on the shipment, such as weight, number of boxes or product type. For security purposes, information on parties involved with the cargo might also be verified. This exchange of data along the entire logistics or supply chain required a new way of thinking about information exchange. A solution was introduced by David Hesketh of of UK Customs and Frank Heijmann of Dutch Customs in the course of the EU-FP7 project INTEGRITY: the data pipeline, see Klievink et al. (2012). This is a construct in which data is collected from all parties involved in the chain, and that data is made available to all parties involved in the chain. While this sounds rather simplistic, such a full information access is not common practice in logistics and supply chain management today. A graphic depiction of the data pipeline concept in a logistics chain is shown in Fig. 7.6. Having full access to information solves several important problems. First of all, it provides a clear answer to the question “who packed the box?”. This is a critical question for any receiving customs agency, because it marks the so-called consignment completion point, which is the moment in which all relevant data on a specific shipment is fixed (see Hesketh (2010) for more discussion). It is this data that should be on the documents, and that functions as a benchmark for easy verification of the goods.

Fig. 7.6 The data pipeline

156

A. W. Veenstra

Second, it provides an identification of all parties in the chain. In practice, many parties receiving goods do not know who was involved in each stage of the chain. Knowing this adds considerably to the faith a company and customs might have in the level of control in the chain. Third, customs might possibly have access to more data than the declaration if there is such a data pipeline in a logistics chain. In this case, questions customs may have and that are cumbersome to answer (who is the real consignor and consignee?, provide us with the commercial invoice!) are easy to satisfy. 7.4.2.3

Becoming a Trusted Tradelane

The previous two subsections have discussed important prerequisites for the trusted tradelane category that was introduced in the new Dutch Enforcement Vision (see Fig. 7.5). This section will propose a recipe for becoming a trusted tradelane. First of all, we start with a list of attributes of trusted tradelane. These attributes have developed out of projects such as the EU-FP7 projects INTEGRITY and CASSANDRA (see http://www.cassandra-project.eu) and the Dutch TRANSUMO project PROTECT (see http://www.transumofootprint.nl, search on “PROTECT”), and are based on many discussions with business parties and customs. A trusted tradelane is: • • • • •

... a chain consisting of known parties, ... with controlled packing of loading units, ... verification of the document flow against the physical flow, ... tracking of the status of goods and transportation, and ... sharing of data along the entire chain.

Obviously, a trusted tradelane cannot be a chain where some of the parties involved are unknown. That would be an unacceptable risk for the operational performance of the chain and it would introduce vulnerabilities for crime and terrorism. We argued above that the packing of the box at the beginning of the chain is a crucial moment that needs to be controlled explicitly. Moving goods creates different flows of information and physical goods (and money). At some point in the chain, the association between these flows must be verified. Customs now often plays this role, but chain partners can also execute this task themselves, and many can probably do this at a better stage in the chain, and at a better time than customs. Tracking and tracing of goods, loading units and vehicles is also an obvious prerequisite. Finally sharing all operational data along the chain reflects the idea of the data pipeline from the previous section. We consider all these conditions to be necessary conditions. They are not sufficient, however. Many things can still go wrong. First of all, as a result of differences in operational procedures and risk perception. Parties may verify the packing of the box, but if they are sloppy, and many mistakes are made, then it is inevitable that some of these mistakes transfer to the receiving end of the chain. Therefore, there has to be a common operating procedure in the entire chain, that avoids mistakes and errors, and that enforces a high standard of operation among all chain partners.

7 International Trade, Global Supply Chains and Compliance

157

Second, there has to be a joint approach to risk that is shared by all parties, ànd customs. This will ensure that standards of operations and verification are adopted in the chain, and are executed diligently by all partners. This may be far from easy. In retail chains, where containers are shipped by factories in manufacturing countries such as China, Vietnam, Thailand, by an annually changing set of contract manufacturers, the introduction of a common risk standard is virtually impossible. Each additional demand made by a retailer or purchasing organisation will be translated into a high purchasing price, which means that risk control translates into product price increases. A third condition is that monitoring and alerting of problems should be pro-active, instead of re-active. There will always be problems in the international movement of goods, because the process is so complicated and involves many parties. If these problems are only identified after the fact, however, for some of them it will be too late to execute mitigating measures successfully. It would be much better to anticipate these problems, wherever possible, and alert other parties in the chain that are in a position to mitigate these problems. Preferably, this mitigation should be transparent, so that the alerting party but also customs can verify what happened. Finally, the combination of pro-active alerting, transparent mitigating measures, open re-alignment of risks, and joint analysis and auditing of events and procedures will result in building trust in the entire chain between business parties and between business and customs. From all these conditions, a procedure can be derived for companies to become a trusted tradelane. In this procedure, the relationship between the companies involved and customs plays an important role. The identification of the steps in this procedure is based on discussions with a number of multinational companies based in the Netherlands that have indicated an interest to obtain, with their chain partners, this trusted tradelane status. The procedure has the following steps: 1. Perform a supply chain mapping process, and describe the current monitoring and alerting approach, 2. Initiate communication with customs on the desire to pioneer the trusted tradelane regime, 3. Continue an active dialogue with customs to establish a joint approach to risk assessment, analysis and mitigation, 4. Build the trusted tradelane status into commercial relationships along the chain. The first step seems simple but mapping supply chains, especially for large multinational companies is not an easy task. The simplest is the identification of parties and operational steps. It becomes more complicated if IT systems, quality procedures, monitoring and alerting are involved. The purpose of this step is to verify to what degree the supply chain setup adheres to the necessary conditions. The second step is crucial, because the trusted tradelane concept is essentially a construct introduced by customs. Involving customs in an early stage is therefore important. Customs also has an important role to play, and not just in the verification of procedures. The international movement of goods involves multiple jurisdictions, and therefore the development of trusted tradelanes will require the involvement

158

A. W. Veenstra

of more than one customs agency. A complete trusted tradelane will not emerge if customs at one end of the chain cannot engage the relevant customs agency at the other end of the chain. In the Netherlands, because of the existence of bilateral trade agreements, countries such as Japan and South Korea are good candidates to develop trusted tradelanes with. With other important countries such as Turkey or the USA this will be more difficult, because these are covered by trade agreements (or trade agreement negotiations) with the EU. The third step is the core of the trusted tradelane process, i.e. the active dialogue with customs on the risk management approach. The approach of customs was described above. Customs will want to gain insight in the totality of risk identification, assessment, and mitigation. In the cases that we are aware of, this is done by the development of a so-called risk matrix. This matrix lists all risks, and describes in detail what these risks are, how these risks are identified, measured, and mitigated, and further, who owns the control procedure for the risk, and how often it is audited. Customs will then infer from the matrix what residual risk remains, and therefore, what additional supervisory measures customs still needs to undertake. The fourth step solidifies the trusted tradelane in all commercial contracts between partners in the chain. In this way, the risk management process is codified in operational contracts and service level agreements. Case Study: Developing Trusted Tradelanes for a High Tech Company The Netherlands hosts a number of top high tech companies, that manufacture complicated components or machines, and serve a global market. From a compliance point of view, these companies are covered by the regular customs export related regimes (see Sect. 7.3.2), as well as by special so-called export controls. These are rules that are designed to prevent the export of intellectual property, and of dual use goods. Dual use goods are goods that have dual-civil and military-use. Examples are lasers, lenses, chips, and so on. These types of goods are separately controlled and this leads to another layer of supervision by customs. For companies, complying with export control regulation is extremely important, because losing the export license will mean the company is, at the very least temporarily, out of business. High tech companies regularly ship consignments of (very) high value, and because of the sensitive nature of these consignments, they are very carefully packed, sometimes in clean rooms. In most cases, international transportation is done by air. It may occur that customs agencies decide that they need to verify the content of the packages against documentation at the airport. This does not happen often, because of the high degree of compliance that these companies already have. But if such a decision is made, it is disastrous. Opening up a carefully packaged high tech piece of equipment usually means that the consignment cannot travel further. It cannot be easily repackaged, especially if packaging was done in a clean room. Inspection in this case effectively means destroying this shipment. Clean rooms, being expensive facilities themselves are usually planned to full capacity, which means that repacking can also result in serious delays in delivery of the product to a client.

7 International Trade, Global Supply Chains and Compliance

159

One high tech company engaged in a dialogue with customs in an attempt to avoid any physical interference of customs during transportation of its goods. The company and customs therefore jointly developed a comprehensive risk matrix that lists all relevant risks that might result in compliance problems for the transportation of high value equipment. This matrix listed several dozen risks, with a prioritization for customs, control measures, monitoring and mitigation procedures and maintenance procedures. Relevant risks include exporting without an export license, incomplete declarations, incorrect master data, and reconciliation problems in the administration of the bonded warehouse. The matrix results in the identification of residual risks, which are small, but not non-existent. In addition, the control by customs of the “packing of the box” in the packing areas is impossible because unannounced entry of customs officials in these facilities may disturb special atmospheric conditions. The big question now is whether this joint analysis of risk is sufficient to completely avoid physical inspections at airports. The risk matrix still leaves residual risk, while the company expects a 100% exemption from physical inspections at the airport. On the other hand, this is one example of a very advanced joint identification and assessment of risk by a company and customs, that is at the same time transparent. The risk management approach is not only clear for customs, but also for the final customs authority in Brussels.

7.5 Concluding Remarks This chapter deals with the international dimension of doing business. We discussed the way businesses organize their global presence in supply chains, and what kind of regulatory regimes apply to the international movement of goods within these supply chains. We have introduced elementary concepts such as arm’s length trading and incoterms that structure the transactions for the movement of goods across borders. At an elementary level, we have positioned the mechanism of international trade in supply chains. We have extended the understanding of compliance in international supply chains by elaborating on global security regimes and the generic principles of customs regulation in the world. In the supply chain management literature, these principles are not well known. We have introduced the new Dutch vision on enforcement, which suggests that a supply chain, or at least a part of it, could be constructed as a trusted tradelane. This reinforces again the relationship we describe between compliance and supply chains. As the most advanced part of this chapter, we have discussed the construction of trusted tradelanes in practice. Important components of this concept are a risk management approach and a solution to share data along the entire chain. We finish the discussion with a conceptual four step approach that businesses can follow to develop

160

A. W. Veenstra

trusted tradelanes themselves. A business case offers some insight in dilemmas that result from the development of (elements of the) trusted tradelane concept. While the first two parts of this chapter are required for a general understanding of the topic of compliance in supply chains, the last, more advanced part, describes very much an ongoing process. Both businesses and customs are currently developing their understanding of compliance management and are seeking pragmatic solutions for the development of trust based compliance concepts. This process is hampered to some extent by the introduction of the new Union Customs Code in the European Union. For the more advanced countries, such as the Netherlands, the consensus is that this Customs Code is in many ways a step back from the previous Customs Code. A number of the advanced ideas about the recognition of compliance and risk management approaches of businesses and translation into less intrusive inspection by customs require reformulation and rethinking. An important aid in this process is more high level, strategic education in this field of customs and supply chain compliance, for both customs professionals in business and customs officials. This chapter is aimed to play a role in that process.

7.6 Further Reading A nice introduction in the process of global operations management in global supply chains is presented by Dornier et al. (2008), enriched with a large number of case-studies from Europe, the U.S., Latin America, and Asia. A recent source on international trade is Brooke and Buckley (2016), while Arntzen et al. (1995) provide insight in the design of a worldwide operating supply chain for a global electronics manufacturer (DEC). A discussion on security in global supply chains is presented by Sarathy (2006). Supply chain risk management is discussed in e.g. Curkovic et al. (2013) in the context of a general framework, while further challenges are elaborated by Manuj and Mentzer (2008) and with respect to the threat of terrorist acts in particular by Sheffi (2001). With respect to trusted tradelanes, a nice example is also the case of Royal Flora Holland, discussed in Chap. 25. Acknowledgements We would like to thank Frank Heijmann from Dutch Customs, Ruud Tusveld from PWC, our fellow teachers in the Master Customs and Supply Chain Compliance and our students for constructive discussions on parts of this chapter.

References Anderson S, Shannon W, Daly J, Marilyin F (1999) Why firms seek iso 9000 certification: regulatory compliance or competitive advantage? Prod. Oper. Manag. 8(1):28–43 Arntzen B, Brown G, Harrison T, Trafton L (1995) Global supply chain management at digital equipment corporation. Interfaces 25(1):69–93 Brooke M, Buckley P (2016) Handbook of international trade. Springer, Berlin

7 International Trade, Global Supply Chains and Compliance

161

Chopra S, Sodhi M (2004) Managing risk to avoid supply-chain breakdown. MIT Sloan Manag Rev 46(1):53–61 Curkovic S, Scannell T, Wagner B, Vitek M (2013) Supply chain risk management within the context of COSOs enterprise risk management framework. J Bus Adm Res 2(1):15–29 Dornier P, Ernst R, Fender M, Kouvelis P (2008) Global operations and logistics: text and cases. Wiley, New York Giermanski J (2012) Global supply chain security. Scarecrow Press Inc, Plymouth Hausman H, Lee H, Napier G, Thompson A, Zheng Y (2010) A process analysis of global trade management: an inductive approach. J Supply Chain Manag 46(2):5–29 Hesketh D (2010) Weaknesses in the supply chain: who packed the box? World Cust J 4(2):3–20 Jüttner U, Peck H, Christopher M (2003) Supply chain risk management: outlining an agenda for future research. Int J Logist: Res Appl 6(4):197–210 Klievink A, van Stijn E, Hesketh D, Alderwereld H, Overbeek S, Heijmann F, Tan Y (2012) Enhancing visibility in international suppyl chains: the data pipeline concept. Int J Electron Gov Res 8(4):14–33 Larson P, Halldorsson A (2004) Logistics versus supply chain management: an international survey. Int J Logist: Res Appl 7(1):17–31 Lee H, Silverman A (2008) Renault’s Logan car: managing customs duties for a global product. Stanford Business School case pp GS–62 Manuj I, Mentzer J (2008) Global supply chain risk management. J Bus Logist 29(1):133–155 Norrman A, Jansson U (2004) Ericsson’s proactive supply chain risk management approach after a serious sub-supplier accident. Int J Phys Distrib Logist Manag 34(5):434–456 Sarathy R (2006) Security and the global supply chain. Transp J 45(4):28–51 Sheffi Y (2001) Supply chain management under the threat of terrorism. Int J Logist Manag 12(2): 1–11 Simchi-Levi D, Simchi-Levi E, Kaminsky P (2007) Designing and managing the supply chain: concepts, strategies, and cases, 3rd edn. McGraw-Hill, New York Viner J (2014) The customs unions issue (Edited by P. Oslington). Oxford University Press, Oxford

Part III

Overarching Topics

Chapter 8

Information Technology J. Rod Franklin

Abstract It is hard to envision the rapid globalization of industry and supply chains without a similar rapid advancement in information and communications technologies (ICT). The increase in organization size and complexity that occurred as businesses expanded their operations across the globe required ever more powerful computer systems to effectively manage operations. Similarly, the expansion of business supply chains needed more advanced communications technologies to enable organizations to manage them properly. With each advance in ICT and each expansion of business operations, the tight relationship between ICT and business growth has become even more important. This chapter provides an introduction to the various systems and technologies that are being used to manage modern global businesses and their supply chains. Since there are myriad different systems and technologies available to industry today, a single chapter does not provide sufficient space for an in depth discussion of any of them. What the chapter does attempt to do is introduce the reader to the various technologies that are currently being used by industry to manage their extended business (basic and advanced sections) and provide insights into some of the new and exciting technologies that will play an important role in the interaction between ICT, business and society in the future (state-of-the-art section).

8.1 Introduction It would have been extremely difficult for businesses to have established complex global operations as rapidly as they did without a concurrent rapid advancement in information and communications technologies (ICT). The increase in organization size and complexity that occurred as businesses expanded their operations across the globe required ever more powerful computer systems to manage operations effectively. Similarly, the expansion of business supply chains needed more advanced communications technologies to enable organizations to manage them properly. With J. R. Franklin (B) Kühne Logistics University, Hamburg, Germany e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_8

165

166

J. R. Franklin

Fig. 8.1 Growth in world exports mapped to trade liberalization regime (World Trade Organization)

each advance in ICT and each expansion of business operations, the tight relationship between ICT and business growth has become even more important. The growth in world trade since the Second World War has been impressive (Fig. 8.1). This growth has been encouraged by the reduction of trade barriers during successive rounds of negotiations by the world’s leading producer countries. However, without a similar set of advances in information and communications technologies it is hard to conceive of this growth occurring as quickly or on the scale that it has. The modern computing age did not begin until after the Second World War. It was not, in fact, until the late 1950s that large industrial users began incorporating computers into their back office operations. The high cost of computing technologies, the special skills required to run and manage the computing systems, and the lack of standard software to run on the systems made computerization in these early days an expensive business proposition (Ceruzzi 2003). The introduction of IBM’s System/360 in the mid-1960s changed this. With the System/360 companies could buy a small system to meet their current needs knowing that as they grew their investments in software and training would not be wasted. The System/360 was the first mainframe computer that ensured that software that ran on its entry-level systems would also run on any of its range of larger systems (Rosen 1969). While the System/360 revolutionized business computing, it was the continuing increase in computing power that has enabled organizations to leverage technology to grow and expand globally. Moore’s Law, named after Gordon Moore a co-founder of Fairchild Semiconductor and Intel, describes how the number of transistors in an integrated circuit doubles approximately every eighteen months (the law has been

8 Information Technology

167

revised to say that doubling occurs now every two years). This “law” has been remarkably accurate and the ability to build more processing power into smaller packages has facilitated the development of not only more powerful data center systems, but personal systems and, most recently, build intelligence into almost any device (Schaller 1997). In a manner similar to that in which computing technology was advancing, telecommunications technologies were also advancing, although not as quickly. In the late 1960s the United States’ Defense Advanced Research Projects Agency (DARPA) began funding research of its DARPANet, which would become the Internet. Improvement to the operation of the Internet continued through the 1970s and 80s but did not really take off until the Internet was opened to the public with the decommissioning of the National Science Foundation Network in 1995. By 1995, the World Wide Web was in the public domain and companies were only too happy to build commercial applications using the various tools that software companies were now developing for operating over the Internet (Abbate 2000). The Internet revolutionized how organizations could communicate over extended distances. E-Mail, electronic messaging, Voice Over IP and other technologies allowed companies to manage their growing supply chains and participate in the rapid rise in global trade. Telecommunications companies struggled during this same time period to address the issue of mobile communications. Portable radio transmitters had existed since the First World War, but telecommunications in the 1970s looked pretty similar to how it looked in the 1920s. In the late 1970s NTT in Japan introduced the first 1G mobile telephone system. This first generation system was superseded in the early 1990s by two differing second generation systems; the European GSM system and the US CDMA system. In 2001 NTT again introduced the next generation system in Japan launching 3G services. The benefit of 3G systems was their ability to provide high data rates (Farley 2005). It was with the advent of 3G systems that mobile communications came of age and became the final backbone of globalization. Mobile telephony based on 3G technology allowed businesses to coordinate with their operations anywhere in the world without worrying about fixed line requirements. Delivery of data over the mobile network enhanced this communication and enabled the transmission of data without the need for fixed lines. The ability of ICT to transform and disrupt industries, as well as to support global businesses in managing their supply chains, has now matured to a point where one could truly say the digital age has arrived.

8.2 Enterprise Systems (Basic) The complexity of modern business organizations requires business managers to rely on information systems to effectively and efficiently manage their operations. Without the use of information technology, modern businesses would have to rely on large staffs of human information tabulators. Such organizations would be neither

168

J. R. Franklin

Fig. 8.2 Typical categories of modern enterprise systems

very flexible nor responsive. Quality of decisions would be suspect due to a lack of timely information on problems and potential data quality errors introduced through the myriad of human interfaces in the chain of command. Modern enterprise systems divide themselves into three typical categories; customer-focused systems, internally focused systems, and supply chain-focused systems (Laudon and Laudon 2015). While the primary discussion in this chapter will be on enterprise systems found in a typical production enterprise, the reader should be aware that all businesses employ systems that fall into one of these three categories. Whether the enterprise produces automobiles, delivers a service, or performs some extractive activity, they will have customers that they wish to manage, internal operations that need controlling, and a supply chain that must be managed as well (Fig. 8.2).

8.2.1 Customer Focused Systems Customer focused systems attempt to provide an organization with a uniform and consistent view of each customer with which it does business. In addition, these systems provide customers with a single view into the organization itself. These systems, collectively called customer relationship management systems, allow an organization to acquire new customers, enhance relationships with existing customers and develop programs to retain (or “fire”) customers that have been acquired. Customer acquisition is supported through the application of contact management and prospecting tools. Selling efforts are documented and progress towards deal closure tracked through the system. Management support efforts can be enacted at any point in the monitored sales cycle should an opportunity be deemed at risk. In addition, personalized direct marketing activities can be employed to attract potential new customers and promote new products. Enhancing relationships with existing customers is a critical part of any organization’s success. A primary objective of any sales organization is to gain a larger share of

8 Information Technology

169

a customer’s spend on products that the organization sells. Relationship enhancement occurs by providing existing customers with tools and services that make dealing with an organization easy. Customers who find that an organization’s after sales support, account management and solutions approach friendly and value-focused become repeat customers. Upselling and cross-selling opportunities are enabled when a customer is pleased with the services provided and the value created through their interaction with the organization. Customer retention is a critical success factor of every organization. The acquisition cost for a new customer sale can be on the order of six times as high as a similar sale to an existing customer. For this reason alone it makes sense to ensure that good customers remain loyal and do not leave. Customer retention systems assist organizations in determining who the valuable customers are. This allows an organization to focus customized programs on these customers and enhance their relationship with the organization. The customer retention system also assists in identifying non-profitable customers and facilitate management actions that either turn these customers into profitable customers or removing the customers from the organization’s customer portfolio.

8.2.2 Internally Focused Systems While the successful management of an organization’s customers is a key success factor, without the ability to effectively and efficiently manage the firm’s internal operations customers will soon leave for better run organizations. Enterprise resource planning (ERP) systems consist of applications that focus on managing each of the internal functions of an organization. The current usage of the acronym “ERP” generally implies an integrated modular system built around a single database enabling a single “true” view of an organization’s operations. In fact, all organizations, whether using a modern ERP system from SAP, Oracle, or Microsoft, or a collection of separate systems, employ “enterprise resource planning” systems. This fact arises because no complex organization can manage its operations without employing functional systems that are integrated in some manner for managerial decision-making. ERP systems are employed to both plan and manage the resources of the organization. The functions that an ERP system usually covers includes the following: • • • • • • • • •

Human resource management Financial management Controlling and accounting Sales forecasting and demand planning Purchasing and supply management Manufacturing planning and scheduling Capacity planning Transport and distribution planning Inventory and warehouse management

170

J. R. Franklin

• Order management. Data from each of these functional activities is collected in a central database so that management can obtain reports on how well the organization is meeting its obligations. While an ERP system provides general planning and management services for the enterprise, the reader should understand that unique functional systems are required for many activities that an organization performs. These unique systems vary by the type of business that an organization competes in. For example, an airport employs air-traffic management systems to manage the inbound and outbound flow of aircraft. These unique systems feed into the airport’s ERP system providing the airport with information on which aircraft have landed and taken off, what services were provided to the aircraft by the airport, etc. These data are used by the airport’s ERP system to bill airlines, schedule workers, plan procurement activities, etc.

8.2.3 Supply Chain Systems An organization’s supply chain can be a vast international web of organizations working on various elements of the firm’s products. Coordinating this web of suppliers, all of whom are working on orders for multiple other organizations, so that a company’s customers receive their orders on time, in the quantities desired, at the quality expected, and at the cost agreed to is an extremely difficult task. The systems that have been developed to address this task have generally grown up independent from traditional ERP systems and utilize advances in telecommunications and the Internet to facilitate what has become known as supply chain management. Supply chain management systems address issues associated with finding reliable suppliers (strategic sourcing), forecasting demand and translating demand forecasts into supply requirements, distribution management, customer order fulfillment, warehouse management, inventory control, transportation management, export/import document generation and control, tracking and tracing in transit goods, managing risks and numerous other activities a firm might engage into supply its customers with products. Each of these areas will be discussed briefly in the sections that follow:

8.2.3.1

Strategic Sourcing

Identifying and managing suppliers to ensure that they deliver the required goods under the contractual conditions that have been negotiated is a critical component of modern supply chain management activities. Systems that have been developed to manage and review supplier performance include performance scorecards, contract management systems, benchmarking systems and negotiation systems. Compliance management and auditing systems are also important elements in managing suppliers

8 Information Technology

171

due to the negative impact on an organization’s reputation if a supplier fails to live up to the compliance standards of the contracting firm.

8.2.3.2

Demand Forecasting

Accurate demand forecasts are critical for the proper management of product flows, customer order fulfillment and inventory levels. Sophisticated forecasting systems, integrated with an organization’s ERP system and customer inventory and point of sale systems, can provide an organization with forecasts that minimize stock outs and reduce capital tied up in inventory. Today these forecasting systems are generally linked to a formal sales and operations planning system/process in which the forecasts are reviewed, enhanced through sales knowledge and used drive operational production and supplier plans.

8.2.3.3

Distribution Management

Systems used to manage an organization’s distribution operations include warehouse management systems, yard management systems (for tracking vehicles delivering goods or picking up goods from the distribution center), dock management systems (for scheduling inbound and outbound vehicles to distribution center docks), value added production systems (for performing value added activities such as product configuration, co-packing, country homologation, kitting and other value enhancing activities), asset management systems for managing forklifts and other material handling equipment, and automation management systems for any automated storage and retrieval (ASRS) or conveyor system that might be employed in the distribution operation.

8.2.3.4

Order Fulfillment

Order fulfillment systems have become increasingly complex with the rise of e-commerce. These systems provide information on available to promise inventory, best inventory for fulfillment (which warehouse should be used to fulfill which order), pegged order priorities (an order is “pegged” once it has been promised, the pegging can be prioritized so that best customers receive orders first or based on other rules that the organization establishes), delivery scheduling and tracking, and other information that might be of use to a customer concerning their orders.

8.2.3.5

Warehouse Management

Warehouse management systems have been discussed previously in the text. However, it should be noted that as warehouses become more than simple storage facil-

172

J. R. Franklin

ities, these systems have become increasingly more like production systems than traditional warehouse management systems. Employee productivity and scheduling modules, pick optimization routines, light production scheduling, packaging, receiving and putaway, shipment management, dock management, yard management, asset management and even environment management are all system modules that various vendors provide today to help manage warehousing operations. In addition, automation of various functions with pick-to-light, pick-to-voice, pick-to-picture, etc., are beginning to become more common in these systems as increased productivity in the picking operation is pursued.

8.2.3.6

Inventory Management

Inventory management systems are generally embedded into an organization’s ERP system. However, systems that add functionality to standard ERP based have been developed to manage things like supplier production constraints (constraint management systems), dynamic relocation of inventory based on demand, end-of-life inventory, repair and spares inventory management, and other aspects of inventory control that go beyond the standard services offered by an ERP inventory management package.

8.2.3.7

Transportation Management

Transportation management systems have been increasingly expanded to allow organizations to better manage the movement of their goods. These systems allow organizations to dynamically develop optimized routes for goods deliveries based on current orders, track the vehicles delivering goods, manage time slots for goods deliveries, track driver driving times, and numerous other parameters concerning the transport of goods. Companies that own their fleets also employ fleet management systems that track vehicle parameters such as operating hours, distance traveled, maintenance and, if the vehicle has its own telematics system, parameters associated with engine operation, acceleration, speed, etc.

8.2.3.8

Export/Import Document Management

Companies that ship goods across international borders must ensure that these goods carry with them the proper export and import documentation. Without proper documentation, export or import authorities will hold the goods until the proper documentation can be delivered. Delays such as this can be costly and lead to extra charges for storage as the goods await clearance. In addition, customer service suffers from such delays. Export/import documentation systems provide information on goods being shipped such as goods category, tax category, whether the goods are consid-

8 Information Technology

173

ered hazardous or controlled, country of origin (based on the importing country’s definitions), whether the goods producer is a known producer, etc. The databases containing all the various country requirements for exporting and importing goods are extensive and require continual updating to ensure that goods are not held at a country’s border (cf. Chap. 7).

8.2.3.9

Tracking and Tracing

Answering a customer’s question concerning where their order is while it is in transit is not a simple task. To answer such a question an organization needs to know which container, which transport vehicle and which location the article is in. These are not easy questions to answer so responding to a simple query concerning the location of an item is difficult. The systems that are used to answer such questions include bar code or RFID systems to identify and locate the item, transport operator vehicle tracking systems, and an event monitoring and processing system to continuously update the organization’s tracking system.

8.2.3.10

Risk Management

Supply disruption has become a significant problem to organizations that have global supply chains. Unplanned events, such as a tsunami, volcano eruption, or hurricane can result in significant monetary loss for organizations that have their supply of critical goods disrupted. New systems, such as DHL’s Resilience 360 software, enable companies to define the risks that they want to track and to actively monitor potential disruptions to their supply chains.

8.2.4 Electronic Data Interchange Because most of the activities managed in a supply chain occur remotely from the organization, supply chain managers rely heavily on modern telecommunication infrastructures and the Internet to perform their tasks. Electronic data interchange (EDI) is a key enabler of supply chain management activities. EDI services involve the electronic exchange of documents between trading partners. EDI standards, developed through the United Nations/Electronic Data Interchange for Administration, Commerce and Transport (UN/EDIFACT) standard provides standard formats for exchanging information between trading partners. EDI messages can be exchanged between partners over the Internet, through direct network-to-network linkages and through value added network service providers. EDI transactions cover the full range of information exchange requirements between trading partners including notification of delivery, inventory changes, invoicing, ordering, pre-advice on delivery, noti-

174

J. R. Franklin

fication of departure, etc. The EDI EDIFACT standard is recognized globally and covered by the International Standards Organization under the standard ISO 9735. Traditionally, EDI messages were communicated between business partners in a structured manner over private and/or value added networks. These networks were closed and required partners to subscribe to the network and map their internal data to the EDIFACT transaction standards of the messages being exchanged. This requirement for subscription to a service and message mapping has a cost associated with it (both monetary and technological) that blocked many smaller organizations from using electronic data exchanges for passing information to their trading partners. As the Internet became more established as a means of communication between partners, and the eXtensible Markup Language (XML) became more widely used, the EDIFACT standard was revised to allow for EDI messages to be transmitted using standard XML templates. This revision of the standard has facilitated greater adoption of EDI by smaller organizations and improved considerably their ability to effectively and efficiently communicate with their trading partners.

8.3 Enterprise Systems Infrastructure (Basic) To support the operation of an organization’s enterprise systems requires significant computing hardware, storage systems, networking systems and telecommunications systems. These systems operate in conjunction with one another through computer operating systems, control systems, communications protocols and security systems. The orchestration of these hardware and systems elements is the job an organization’s management information systems (MIS) department. While the number of systems that need to be tied together in a modern global enterprise appears to be rather large, MIS departments have developed architectures that facilitate both the transactional requirements of operational departments with the managerial decision making requirements of managers. These enterprise systems architectures link operational systems to centralized data and reporting systems through enterprise application integration (EAI) buses. Applications or databases requiring information from various operational systems “subscribe” to the information that the operational system “publishes.” The “pub/sub” concept has proven extremely flexible and successful, particularly when integrated with the EAI bus’s extract, transform and load (ETL) functionality. This functionality allows each operational system to encode its information in a format that it can use while providing a translation service for other applications that need the data, but use a different format. A common structure of such an architecture appears in Fig. 8.3. The physical infrastructure that is used to manage an organization’s systems usually resides in a corporate data center. These centers incorporate gateways, switches, servers, management software and backup capabilities configured in such a manner as to provide high systems uptime at low operating costs. A typical configuration of a turn of the century data center appears in Fig. 8.4.

8 Information Technology

175

Fig. 8.3 A typical application architecture for a modern organization

Fig. 8.4 A turn of the century data center (Courtesy IBM)

Data centers, like the one shown in Fig. 8.4, are operated on a 7 × 24 × 365 basis with an uptime expectation of almost 100%. This means that data center managers must ensure that there is sufficient capacity to handle peak loads without response time degradation. They must ensure that there is backup and mirroring of software and data so that smooth fail overs occur if a fault arises in either a running program or a piece of hardware. The data center managers must also ensure that there is a high level of security implemented in their operations to avoid system intrusion by

176

J. R. Franklin

unauthorized individuals. These managers must perform these functions in a costeffective manner and one that minimizes environmental impacts. The physical systems used in a modern data center have changed dramatically since the beginning of this century. It would not have been unusual at the turn of the century for a data center to be composed of a single mainframe computer with traditional Winchester disk drives arrayed in storage clusters around the mainframe. Management would have looked at capacity expansion and storage expansion as major investments that added blocks of capacity when processing or storage constraints arose. The cost for upgrading capabilities in these data centers meant that they were inflexible when unplanned demand growth arose. The cost of increasing capacity in the data center, coupled with the poor perception that MIS departments carried because of the data center’s inflexibility created significant tension between data center managers and the rest of the organization. Today, a data center is quite different from this old model (Fig. 8.5). Technological advances in processing capabilities, storage systems, inter-data center communications and infrastructure management systems have revolutionized how the modern data center is structured. Instead of single mainframes or several distributed mainframes, current data centers use racks of inexpensive processors and flash storage devices. The processing of operations is distributed across the many processors through software enabled “virtualization,” a process in which workloads are assigned to “virtual machines” and controlled by a high-level software controller called a hyper visor. This approach to managing data centers has evolved out of the needs of organizations such as Google where high flexibility in capacity, ease of expansion and cost

Fig. 8.5 The interior of a modern data center showing racks of blade processors (Courtesy Microsoft)

8 Information Technology

177

efficient operations created demands for a new data center model. This model for creating and managing data centers has also facilitated the emergence of the cloud computing paradigm, the subject of the next section of this chapter.

8.4 Cloud Computing (Advanced) Cloud computing has evolved rapidly to become one of the most disruptive technologies to impact enterprise computing since the advent of the IBM System/360 computer. Because of its rapid advance, there are many perceptions (and definitions) of what cloud computing really is. In an attempt to address these competing viewpoints on cloud computing, the National Institute of Standards and Technology (NIST) in the United States established an “official” definition of what it considers to be cloud computing (Mell and Grance 2011). Cloud computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.

The NIST definition of cloud computing includes five essential features of any cloud computing environment. These features are: 1. On-demand self-service: computing capabilities can be provisioned at any time by users without the intervention of anyone at the cloud service provider; 2. Access from anywhere: services are provided over the Internet or private network in such a manner that they can be accessed anywhere, at any time from any device; 3. Resource pooling: the service provider’s resources are pooled (virtualized) to serve multiple customers in a multi-tenant model of computing; 4. Rapid elasticity: resources can be automatically scaled upwards or down depending on demand; 5. Measured services: services are monitored for usage and fees are charged based on what is used. Services are provided by cloud computing service providers in one of three different service models. Software-as-a-Service (SaaS) is a service model in which users access a vendor’s software that is running on a cloud platform (e.g., Salesforce.com). Platform-as-a Service (PaaS) is a service model where users build their own cloud applications using development tools and infrastructure of a cloud service provider (e.g., Amazon Web Services). Infrastructure-as-a-Service (IaaS) is a service model where the cloud service provider provides hardware and management services to users who then use their own tools to develop and manage their cloud based applications. Cloud service providers can provide public clouds in which users share the cloud platform with other users (e.g., Microsoft Azure). Clouds can also be private where an organization builds its own cloud infrastructure and applications for use by internal

178

J. R. Franklin

Fig. 8.6 NIST standard definition of cloud computing

resources or for sale to external entities (most large public cloud providers have private services, e.g., Microsoft, VMware vCloud, OpenStack, Platform9, and Apache all provide private cloud services). Hybrid clouds provide cloud services to a closed community of users. These types of clouds are public for community members, but private if a user is not a community member (Fig. 8.6). Cloud computing provides significant cost benefits to users by leveraging scale economies and using a “by the drink” approach to charging. Estimates have indicated that infrastructure costs in a normal data center take up approximately 50% of an IT organization’s budget. Cloud economies can reduce this figure by up to 80% leaving a significant amount of “new” money available to an IT organization for application development and enhancement. Almost all major software application companies are now targeting the cloud for deployment of their software. SAP, Oracle, Microsoft, Apple, to name only a few, are pushing their primary applications to the cloud to facilitate growth in scale while reducing overall costs. The platforms that these software companies are creating provide users with enhanced services such as access from any device anywhere at any time. An example of how such a platform for logistics services was developed under research funding from the European Union appears below.

Fig. 8.7 FIspace—a multi-domain collaboration space for business

8 Information Technology 179

180

J. R. Franklin

8.5 Case Study: The FIspace Cloud Based Collaboration Platform The ultimate aim of the FIspace project was to develop, validate, and establish a future business collaboration space (the FIspace) that facilitated radical improvements for information exchange, communication, and coordination among business partners. The FIspace platform used advanced Internet technologies and was implemented in an open manner so that other European Union research projects, as well as external IT providers and interested users, could easily use, test, and exploit its features and services and contribute to its expansion and establishment. Figure 8.7 depicts the overall vision for the FIspace service. The FIspace vision was to become a value added Collaboration Space in the Cloud that enabled actors operating in Collaborative Business Networks in various application domains to seamlessly interact, communicate, and coordinate activities with business partners and to easily create and act in open and dynamic networks of connected businesses. The FIspace propagated a future business model for enabling the rapid development of high-quality ICT solutions at minimal costs by enabling the provisioning, consumption, and re-use of ondemand solutions in the Cloud. General business, as well as domain-specific, functionalities (referred to as ‘Apps’, as the envisioned usage and economic model is similar to mobile apps for smartphones) are developed by IT solution providers (beneficiaries and external providers). These ‘apps’ are provided via the FIspace Store, from which the Apps can be consumed and new Apps can be developed by re-using the features of existing ones. The Apps are selected based on a number of criteria including their functionality, pricing model, past reliability, focus, etc.; furthermore, the Apps can be “mashed up” for individual business needs using the mechanisms and tools provided by the FIspace; this allows for the rapid creation of integrated solutions, composed of multiple Apps, that address specific B2B requirements at minimal cost. In this way the FIspace enables businesses to proactively act on issues or business opportunities without having to incur the overhead and cost that traditionally plagued monolithic applications. The FIspace pushes boundaries on how business software works, facilitating innovation and market impact by laying the foundation for adoption by large user groups and external solution providers that can provide additional, novel, and disruptive Apps for the FIspace. An important point to note is that the innovative aspect of the FIspace model is the model itself (covering both the technical solution and the business model)—it is not any specific ‘technology innovation’ such as, e.g., new algorithms, software engineering concepts or the like.

8 Information Technology

181

8.5.1 The FIspace: Main Features and Building Blocks The FIspace is a SaaS application for facilitating seamless, efficient, and effective business collaboration across organizational boundaries and facilitating the establishment of ecosystems with business benefits for both users from industrial sectors as well as the IT industry. It builds on concepts that have been designed and pre-validated in Phase 1 use case projects, particularly the ‘Collaboration and Integration Platform for Transport and Logistics’ from the FInest project (See Stollberg 2012); the FIspace extends this work towards a multi-domain business collaboration space, and adds the software infrastructure supporting the future business model for FIspace Apps. In addition, desirable domain-specific capabilities are developed, e.g., for inter-organizational process coordination, operational monitoring and tracking, event-driven replanning, ecosystem application development, monetization, and security and privacy management in business networks. In order to realize the envisioned features—i.e., (a) the future support for collaboration in business networks, and (b) a future business model for rapid development of high-quality ICT solutions at minimal costs—the FIspace consists of the following primary building blocks (see Fig. 8.7). (1) The Front-End that serves as the main point of access for End-Users and offers a novel ‘all you need in 1 place’ user experience, including the following features: • Customizable End-User Cockpits • Social Networking and Collaboration Features for Business Partners and Communities, and support for seamless collaboration on specific business activities and transactions • Access anywhere via any device (2) The FIspace Store that encompasses an extensible collection of Apps that can be used and combined for the individual needs of End-Users and re-used for the rapid and easy development of new Apps, including: • The software infrastructure to support the provisioning, consumptions, purchase, and re-use of FIspace Apps for both End-Users and App Developers • Financial management of the FIspace (pricing, payment, revenue sharing)

182

J. R. Franklin

(3) A System and Data Integration Layer to allow for the integration and continued usage of existing legacy and business systems and the integration of external systems and services, including support for: • Connecting business and legacy systems used by individual users • Handling heterogeneous data • Connecting external Systems and Services (e.g., IoT systems, third party and public services) (4) A comprehensive framework for Security, Privacy, and Trust Management to ensure the secure, reliable, and trustworthy handling of business data making the FIspace ‘secure by design’ including: • Access Control and Identity Management for all users • A set of Security Mechanisms to ensure information security, attack prevention, etc. • Developer support to ensure correct usage of necessary security mechanisms in FIspace (5) An Operating Environment that ensures the technical interoperability of FIspace Components and Apps and the consistent behavior of the FIspace, including: • Technical Interfaces and Protocols for the FIspace Components and Apps • An Enterprise Service Bus to support the interaction of FIspace Components and Apps • Replication and Consistency Services to ensure fault-tolerance and transaction support (6) A Development Toolkit providing tool-supported techniques for the FIspace, particularly for: • App Developers to support the development and provisioning of Apps in accordance to the technical governance and procedures of the FIspace • Business IT Experts who customize and extend the FIspace to the individual needs of End-Users at an individual or organizational level A central element of the FIspace is the concept of so-called Collaboration Objects that capture the information that needs to be exchanged among collaborating stakeholders along with the status information of the collaborative business activity, constituting a global knowledge base for collaborative business processes. In close connection with an Event Manager that captures and pre-processes both manual (from humans) and automated (from connected systems) events, this ensured that all information and status updates were provided to each involved stakeholder in real-time. Building upon the initial technical

8 Information Technology

183

design developed in the FInest project,1 the FIspace encompassed the necessary software infrastructure for defining and executing Collaboration Objects in an event-driven manner; an initial and extensible set of Collaboration Object templates along with the corresponding Event Handling Rules were defined for early trials. All FIspace Apps use this technology in order to ensure real-time status updates and information provisioning throughout collaborative business networks, such as, e.g., receive real-time information about the location and status of perishable goods during transport, notifications about deviations from planned execution, etc. Moreover, FIspace Apps may use these objects for devising collaboration processes for specific purposes and even create and modify such processes on the fly, especially for handling deviations that require immediate action. Summarizing, the central features of the FIspace are: • An open application that can be extended and customized for specific stakeholder requirements by integrating domain apps (similar to iPhone/Android business models); • An integrated Front-End as a central point of access to all features and apps, enabling a ‘all you need in one place’ user experience with customizable views for individual actors; • Integrated novel technologies that provide the basis for future collaboration in business networks, enabling ‘all information to each stakeholder in realtime’ as well as the seamless interaction and coordination among business partners; • Information integration from legacy and third party systems enabled through a service-based integration layer; • Modern technologies that ensure the secure, reliable, and trustworthy handling of business information in the Cloud, and the consistent operation of the FIspace; and • An (App) Store that facilitates the provisioning, consumption, and marketing of novel on-demand solutions for B2B collaboration, along with an integrated toolkit for developers. The FIspace collaboration platform is just one example of types of novel new software services that the cloud approach to computing enables. Numerous service providers have developed platforms that reside in the cloud for marketplace operations. eBay, Microsoft, Freighthub, Apple, Amazon, Google, SAP and many others have found that using the cloud to host their applications is both an economical and scalable approach to providing software solutions to customers.

1 The preliminary design of these components has been elaborated in work packages WP5 and WP6

of the FInest project, see http://finest-ppp.eu/project-results/deliverables.

184

J. R. Franklin

Because so many different organizations have started to develop proprietary cloud platforms, a need for standards has been identified. In its effort to define what cloud computing is, the National Institute of Standards and Technology in the United States has developed a reference architecture for cloud computing (Liu et al. 2011). The NIST reference architecture defines five primary cloud actors. These actors and their roles are explained below. 1. Cloud Provider: the cloud provider operates the cloud infrastructure. The provider acts as a service orchestrator combining system components to support the provider’s ability to deliver cloud services to its customers. The provider manages their services by providing effective business support activities to its consumers, provisioning and configuring its services to meet changing customer requirements, and migration and interoperability services so that customers can move their existing data to the cloud. 2. Cloud Broker: cloud brokers act as intermediaries between the cloud provider and the cloud service customer. Their role is to add value to the consumer by enhancing the service provision activity and removing administrative burdens from the consumer. 3. Cloud Auditor: cloud auditors provide independent evaluation services for customers concerning cloud provider security, service level performance, and adherence to standards. 4. Cloud Carrier: the cloud carrier is the Internet service provider, telecommunications company, or value added network provider that acts to connect the cloud consumer with the cloud provider. 5. Cloud Consumer: the cloud consumer is the customer of the cloud service. An overview of the NIST’s architecture appears in Fig. 8.8 that follows. The large number of software service providers that use the cloud has created the opportunity to collect massive amounts of data that previously could not be aggregated and analysed. Organizations have also found that by being able to access remote operations through the cloud has allowed them to obtain vast amounts of data that in the past they could not easily access. This ability to easily access data, both from internal systems and external sources, combine it in novel ways and obtain useful insights from it has created additional opportunities for organizations to create strategic value; all enabled through the use of the new computing model of cloud based services and the data centers that support these services.

8 Information Technology

185

Fig. 8.8 NIST cloud computing reference architecture

8.6 Big Data (Advanced) Big data is defined by its volume, velocity and variety, the three V’s. Volume represents the amount of data that is now accessible to organizations and is measured in Terra and Peta bytes. Velocity is a function of the speed at which data is generated, particularly from sensors, video streaming, and other real time data capture processes. Variety denotes the heterogeneous nature of the data where video files can be combined with text files, sensor data and operational data for analysis and decision making. Sensor feeds providing location and environmental statuses, combined with traffic data, weather forecasts, community social data, and myriad other data elements can be combined to generate sophisticated predictive models. These models can assist a company in understanding where delays might occur in the delivery of goods, how markets for products are evolving, where the organization should target its advertising, how consumers purchase and use their products, and even in predicting whether a certain consumer at a certain location will want to purchase one of their products on a certain day. The ability of big data to support the decision making processes in a business is limited solely by the imagination of the organization and its capabilities in the area of data analytics. In an effort to address the many approaches to defining big data and developing processing services to manage big data activities, the NIST has once more attempted

186

J. R. Franklin

Fig. 8.9 NIST big data reference architecture

to establish a standard architecture for big data management. This reference architecture is part of the NIST’s Big Data Interoperability Framework project, which is attempting to provide a standard for the United States Government in the use and application of big data (Group 2015). The NIST reference architecture is shown in Fig. 8.9. As noted in the NIST reference architecture, processing of big data requires collecting and then preparing the data for analysis. In both the steps, the data analyst must understand what data they wish to collect and why. This facilitates the curation of the data so that proper analytics can be performed on it and meaningful insights generated. The reference architecture also points out that data can be collected in batch, interactive or streaming modes. Each type of data collection option requires its own tool set for the proper collection and analysis of the data. Similarly, big data can be stored in different kinds of databases based on its content. File structured databases, relational databases and indexed databases all can be used to store these types of data based on how unstructured the data is.

8 Information Technology

187

8.7 The Internet of Things (State-of-the-Art) The Internet of Things (IoT) is based on a model in which intelligence is embedded in physical objects and these objects can communicate with one another using either the Internet or other radio frequency technologies (e.g., RFID, GSM, etc.). The IoT is expected to grow at an increasingly rapid pace as standards for communication and security are developed and edge based communication protocols are formalized. Because IoT devices will become pervasive it is critical that the information that they generate be managed in such a manner that the Internet, and authorized listening systems, are not overloaded with information that is of limited use. As an example of the potential problem that unconstrained information processing could place on the Internet and an organization’s processing capabilities, it is estimated that a Boeing 787 generates approximately 0.5 TB of data during each of its flights. The vast majority of this data is never used, but the small portion that is used is critical to the safety of the passengers of whichever airline is flying the plane. IoT devices come in many shapes and sizes and communicate via numerous protocols. Industrial IoT devices go by numerous names as well. In the model of Industry 4.0 IoT sensors enable machine-to-machine communications. In the world of intelligent transport, IoT sensors allow infrastructure managers to manage traffic flows, understand the maintenance status of roadways and bridges, provide directions to drivers and provide traffic information to interested parties. As already noted, for airlines, trucking companies, shipping companies, and other organizations that own complex, capital intensive assets, IoT sensors provide information on operational status, environmental loads, and other information that allow the asset managers to optimize the use of their assets. Between vehicle communications is also enabled via vehicle-to-vehicle communications and sensors. The unfortunate problem with this cornucopia of IoT devices and implementations is that their ability to interact in an interoperable manner is limited. The chart that follows shows just how complex this current “tower of babel” is in the world of IoT (Table 8.1). The issue of standard protocols for interoperability is being addressed by a number of different consortia as well as standards bodies such as the ISO. As with any young technology IoT communications protocols will have to be harmonized over time so that the various connected devices can integrate seamlessly without the considerable specialist intervention required today. To manage the continuous generation of data that the billions of IoT enabled devices will create requires a rethinking of the cloud computing model. Because IoT devices reside at the edge of the cloud and generate enormous amounts of data this new model of computing integrates the cloud with the “fog” of data being generated at the edge of the cloud. Fog, or edge, computing requires that intelligent devices, fog nodes, be deployed wherever IoT devices are in operations. These nodes receive the data generated by the IoT devices they are responsible for, analyse this data for signals that managers have indicated are important and communicate to the cloud only those sensor readings that have been identified as important. Depending on the volume of data and processing times required to analyse the data, there can be a

188 Table 8.1 Current protocol landscape of IoT Protocol layer Infrastructure

Identification Transport

Discovery

J. R. Franklin

Protocols 6LowPAN, IPv4/IPv6, UDP, RPL, QUIC, Aeron, uIP, DTLS, ROLL, NanoIP, CCN, TSMP EPC, uCode, IPv6, URIs WiFi, Bluetooth, LPWAN, ZigBee, Ethernet, ISA100.11a, DigiMesh, IEEE 802.15.4, NFC, ANT mDNS, Physical Web, HyperCat, UPnP

Data

MQTT, MQTT-SN, CoAP, SMCP, STOMP, XMPP, Mihini/M3DA, AMQP, DDS, LLAP, LWM2M, SSI, Reactive streams, ONS 2.0, REST, HTTP/2, SOAP, Websocket, JavaScript

Semantic

IOTDB, SensorML, Semantic sensor net ontology, Wolfram language—connected devices, RAML, SENML, LsDL

Multi-layer

Alljoyn, IoTivity, IEEE P2413, Thread, IPSO application framework, OMA LightweightM2M, Weave, Telehash

Security

OTrP, X.509

hierarchy of fog nodes where lower level nodes closest to the IoT devices respond rapidly to IoT signals and aggregator nodes higher up further process the data for final passage to cloud based systems. The key is the creation of filtration process that ensures that only actionable data required at a particular decision making level is passed upwards to that level. The benefits of a world in which intelligence is embedded into every physical object are considerable. The risks are also considerable. Issues associated with device security, privacy issues, and trust related issues all contribute to making a fully connected world one in which it is only a short conceptual distance to George Orwell’s vision, as described in his book 1984, of a world where governments monitor every action of their citizenry. Much work is still required to determine how to leverage the benefits of a fully connected world while mitigating the potential damaging potentials of real-time monitoring of everyone.

8.7.1 The Physical Internet One potential outcome of a more connected world could be the transformation of how people and freight move from points of origin to destination points. If one uses the analogy that shipments or personal movement can be likened to the movement of messages on the digital Internet, then one can envision a world in which intelligent

8 Information Technology

189

physical objects are transported in an automated and seamless manner between points based on specified quality of service, cost and timeliness criteria. This concept of a “Physical Internet” could provide significant benefits by optimizing transport asset utilization, minimizing congestion, lowering carbon emissions, reducing noise and generating numerous other efficiency related outcomes. Intelligent objects, being shipped in standardized intelligent containers, carried on intelligent vehicles over intelligent infrastructures provides one with the ability to envision the shipment of physical objects in a manner that mirrors the movement of digital messages. Similar to the digital Internet, a Physical Internet would operate like a complex network of networks shared by its users. Nodes within the Physical Internet would act to receive, absorb, produce, store, deconsolidate and reconsolidate, and forward shipments onwards. Transport entities would operate over the infrastructure to carry goods onward toward their next or final destinations, where the infrastructure provides intelligent ‘arcs’ between the nodes. The nodes, infrastructure and transport entities would constantly make relevant data accessible, in a controlled manner, to each another to ensure that any problems were transmitted in real time and shipments could be rerouted to maintain delivery commitments at the lowest effort and cost. This is an interesting concept, but requires considerable research into both the technical requirements needed to operate such a network and the business models that would be necessary to ensure that organizations would use the network. In addition, governance mechanisms and asset ownership models would need to be examined to understand how such a network might operate. However, as with many of the other new and disruptive models enabled through the creative use of IoT, it is a concept well worth exploring.

8.8 Further Reading Several standard texts on Management Information Systems provide the reader with a good overview of general hardware and software used in industry today. The texts below have proven useful to the author and should the reader with a good foundation in modern corporate computing. Kenneth C. Laudon and Jane P. Laudon (2015), Management Information Systems: Managing the Digital Firm, 14th ed., Pearson, ISBN-10 0133898164. Ken J. Sousa and Effy Oz (2014), Management Information Systems, 7th ed., Cengage Learning, ISBN-10 1285186139. With respect to telecommunications a very good book to read as an introduction is: Eric Coll (2016), Telecom 101. 4th ed., Teracom Training Institute. Cloud computing has numerous books and articles that provide a good foundation for understanding the approach to computing services. The following two books are good for those wishing a bit of technical as well as management understanding of the cloud.

190

J. R. Franklin

Dan C. Marinescu (2013), Cloud Computing: Theory and Practice, Morgan Kaufmann, ISBN-10 0124046274. Rajkumar Buyya, Christian Vecchiola and S. Thamarai Selvi (2013), Mastering Cloud Computing: Foundations and Applications Programming, Morgan Kaufmann, ISBN-10 0124114547. Big data is a hot topic with lots of books being written to describe various elements of the idea. Many books deal with the ideas of MapReduce, Hadoop, and other approaches to managing the data itself. However, a couple of good introductory books on big data are: Nathan Marz and James Warren (2015), Big Data: Principles and best practices of scalable realtime data systems, Manning Publications, ISBN-10 1617290343. Anil Maheshwari (2016), Big Data Essentials, Amazon Digital Services LLC. The Internet of Things is a very overhyped topic with lots of material being published concerning it. A good introduction to the topic is: Samuel Greengard (2015), The Internet of Things, MIT Press, ISBN-10 0262527731. To learn more about the Physical Internet go to the following web page: http://physicalinternetinitiative.org/. See also Chap. 31 of the current volume.

References Abbate J (2000) Inventing the internet. MIT Press Ceruzzi PE (2003) A history of modern computing. MIT Press Farley T (2005) Mobile telephone history. Telektronikk 101(3/4):22–34 Laudon KC, Laudon JP (2015) Management information systems: managing the digital firm, 14th edn. Pearson Education Limited Liu F, Tong J, Mao J, Bohn R, Messina J, Badger L, Leaf D (2011) NIST cloud computing reference architecture. NIST Special Publication 500-292 Mell P, Grance T (2011) Special publication 800-145. NIST, U.S. Department of Commerce, p 3 NIST Big Data Working Group (2015) NIST Big Data interoperability framework: volume 6, Reference architecture, NIST Special Publication 1500-6 Rosen S (1969) Electronic computers: a historical essay. ACM Comput Surv (CSUR) 1(1):7–36 Schaller RR (1997) Moore’s law: past, present and future. IEEE Spectr 34(6):52–59 Stollberg (ed.) Conceptual design of domain-specific future internet (FI) platform for transport and logistics. FInest Deliverable D3.2. http://finest-ppp.eu/project-results/deliverables. Accessed 31 March 2012

Chapter 9

Actionable Sustainability in Supply Chains Árni Halldórsson

Abstract Supply chain management offers companies a great opportunity to work with all three dimensions of sustainability—profit, planet and people. The level of achievement depends upon how well these three are addressed both up-stream towards the supplier base and down-stream together with customers and consumers (basic). Actionable sustainability requires managers to relate root causes to impact of sustainable performance, to engage relevant with actors, and to determine system boundaries for their actions (advanced). Further, converting goals into actions is likely to require logistics service innovations, which can range from ad hoc to radical innovations (state-of-the art).

9.1 Introduction: Evidence and Sense of Urgency (Basic) Carter and Rogers (2008: 361) state that supply chain professionals can be seen as being in an “outstanding position to impact sustainable practices”. Today, both scientists and headline news agencies provide ever mounting evidence that responsibility for environmental and social impact of operations within and across companies in supply chains should be our highest priority. Operational processes that focus on supply chain performance priorities such as flexibility and speed, often seen as key customer service criteria in supply chains may have a direct and negative impact on climate change and the natural environment in terms of excessive use of nonrenewable resources and carbon emissions. Such strategic priorities, which are often hyped as excellent achievements, may also come at the cost of social conditions and equality. Meeting often unrealistic or unnecessary deadlines created by “someone” in the “wider supply chain” can put workers health and safety at risk. Worse still, there are examples where aggressive cost-focused sourcing practice and the absence of a structured approach to supplier base analysis has enabled the use of Á. Halldórsson (B) Department of Technology Management & Economics, Division of Service Management & Logistics, Chalmers University of Technology, Gothenburg, Sweden e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_9

191

192

A. Halldórsson

child labour up-stream the supply chain. Together, these economic, environmental and social dimensions are commonly referred to as the triple-bottom line (TBL). A brief summary of the past progress and the present state of supply chain practices reveals considerable room for improvement: • First, at the macro-level and a national level, efforts have been made through e.g. legislation, policies and international collaboration to make individual companies more responsible for sustainable development. • Second, supply chain practices such as global sourcing and time-based distribution may be beneficial from the perspective of unit costs (often cheap labour cost and access to raw materials) but may result in unsustainable practices, both social (e.g. working conditions and use of child labour in plants and warehouses) and environmental (e.g. longer transport distances that increase carbon emissions). • Third, related to this is an array of inefficiencies in supply chains, including excessive inventories, poor cargo load factors or even empty travel in freight transport (i.e. shipping air and packaging), products that are stored where not needed (poor customer service levels, lost revenue), products that never sold are returned back up-stream the supply chain due to ambitious or misguided sales plans, or in some cases simply thrown away due to high costs of product returns, and unnecessary movement of products around the world, to name a few. • Fourth, a number of companies have integrated sustainability with their brand and value propositions by making clear statements at the point of consumption (e.g. artefacts and images in retail outlets) about practices that commit the company to sustainable SCM practices (by e.g. illustrating “from bean-to-cup” at a retail counter). Such marketing and public relations efforts may be driven by marketing, but require active involvement of both purchasing and logistics to become true. In this chapter, we present a perspective on sustainability in supply chains with a focus on means by which managers and their organisations can convert visions into actions. By this, sustainability is not reduced to the impact of legislation, policies and technological development but rather seen as managerial decisions and/or behaviour (see e.g. Wu and Pagell 2011). The changes required for sustainable development are a global challenge that require systemic understanding and effort. To this end, the nature of logistics activities and the “supply chain” as a boundary-spanning concept—flows across functions, firms, industries, and regions—offers a suitable level of analysis to understand both root causes and actions to be taken. In order to operationalise sustainability this chapter is based on the following logic. First, we present key concepts and a figure that summarises the notion of ‘sustainability in SCM’. Subsequently, we present an interactive framework in which sustainability—or the triple-bottom line (TBL)—is seen as the result of interaction between suppliers, customers and consumers. To develop a sustainable SCM, new approaches are needed. In combination with four modes of innovation, this will help companies to convert goals into actions, i.e. to go from the principles to the practice of actionable sustainability in supply chains.

9 Actionable Sustainability in Supply Chains

193

9.2 Sustainability in Supply Chains (Basic) This section presents key terms, and how sustainability can be combined with SCM.

9.2.1 Key Terms Supply chain management: Supply chain management (SCM) has developed out of disciplines and professional areas such as operations management, logistics and and purchasing/procurement (often termed supply management, not to be confused with supply chain management). This chapter builds upon the definition of SCM by Lambert et al. (1998: 1) as “the integration of key business processes from end-user through original suppliers, that provide products, services and information that add value for customers and other stakeholders”. Sustainability: An early adoption in logistics and SCM made a strong reference to reduction of freight transport externalities (see overview by McKinnon (2016) and to reverse logistics (Stock 1992)). A hierarchical approach to waste reduction (Sarkis 2003) is common in SCM. Classification of possible environmentally conscious business practices ranges from reduction (e.g. through lean management), to reuse (by using e.g. standardized and reusable packaging materials and containers), remanufacturing (replacement of parts, and sell on secondary markets), to recycling (of materials) and finally disposal (avoid landfill and incineration). Since then, the field has undergone a major change; the view on sustainability has been extended to include an array of functions, ranging from sourcing through production towards distribution of goods and services to the market. Further, logistics has been given a strong interorganisational profile by the introduction of supply chain management (SCM), implying that operational excellence will not only be achieved through internal operations but to a greater extent by collaboration with suppliers and customers. A further extension of sustainable SCM is commonly referred to as triple-bottom line (TBL)—a framework that integrates three dimensions of sustainable performance: social (people), environmental (planet), and economic (profit) responsibility (see e.g. Carter and Rogers 2008). Two common views on sustainable SCM are: …the strategic, transparent integration and achievement of an organization’s social, environmental, and economic goals in the systemic coordination of interorganizational business processes for improving the long-term economic performance of the individual company and its supply chains (Carter and Rogers 2008: 368). …the management of material and information flows as well as cooperation among companies along the supply chain while taking goals from all three dimensions of sustainable development, i.e. economic, environmental and social, and stakeholder requirements into account (Seuring and Müller 2008: 1700).

Both definitions are derived from the triple-bottom line as a perspective on sustainability, and stress the importance of interorganizational engagement with both

194

A. Halldórsson Focal company Supplier base

Sourcing

Production

Logistics

Customers Consumers

Closed-loop supply chain

Environmental (planet)

Social (people)

Economic (profit)

Fig. 9.1 Triple-bottom line across the supply chain

customers and suppliers. Organizations have adopted the TBL framework to evaluate their performance in a broader perspective to create greater business value. The potential of achieving sustainable performance in and through SCM depends on how well the focal company can integrate and mobilise efforts across the horizontal and vertical dimensions of Fig. 9.1. Vertically, Fig. 9.1 encapsulates sustainability in the three dimensions of the triplebottom line: economic (profit), environmental (planet) and social (people). These dimensions reflect the range of options SCM professionals can mobilize to improve the sustainable performance of their organisations. Horizontally, the figure suggests that working with customers and/or suppliers is necessary to mobilise across at least two if not three of the vertical—triple-bottom line—dimensions. Accordingly, the level of sustainability practices relates to the extent of which all three vertical dimensions—people, planet, profit—are addressed, and also to what extent these operate at the horizontal axis of Fig. 9.1, i.e. upstream (suppliers) or downstream (customers, consumers) actors in the supply chain.

9.2.2 Societal and Technological Developments Driving a Sustainable Agenda for SCM Trends create a rhetoric in the organisation and an imaginative vison of the future when goals for sustainable supply chains are decided upon and communicated to stakeholders. Trends relevant to mobilizing a change for sustainable SCM include: United Nations sustainability goals: In 2015, the United Nations presented a new sustainable development agenda, entailing 17 goals set out to “end poverty, protect the planet, and ensure prosperity for all”.1 The following goals will directly affect the agendas of supply chain management professionals: (1) Responsible consumption and production requires companies to assess the resource- and energy efficiency of their operations, and convert their supply 1 http://www.un.org/sustainabledevelopment/sustainable-development-goals/.

9 Actionable Sustainability in Supply Chains

(2)

(3)

(4)

(5)

195

chain strategies so a more sustainable infrastructure (e.g. transportation modes) can be used to produce and deliver products and services to the market place. Another example is design of closed-loop supply chains that can create the effect of Industrial Symbiosis in which companies use each other’s waste or excessive material as input for their own production process. Not only does this reduce use of non-renewable resources, but also reduce transportation costs, and hence energy consumption elsewhere in the supply chain. Climate action refers to measures taken by companies in adapting or changing e.g. speed of delivery processes and requirements through e.g. supplier selection to generate affordable and scalable solutions that balance the various objectives of sustainability (the vertical dimension in Fig. 9.1). The actions may range from incremental changes to radical innovations, spanning processes within and across the boundaries of the firm (the horizontal dimension in Fig. 9.1). Affordable and clean energy and increased use of renewable energy sources (e.g. use of biofuels and natural gas) puts focus on technological solutions in the transport part of the supply chain through clean technology and fuel. Here, the managerial task is about assessing investment options rather than process improvement or re-organising activities, as in the first two goals. Industry, innovation and infrastructure is particularly relevant for transportation in the supply chain, and for the use of digital technology to reduce energy consumption of transportation and/or increase its energy efficiency. Sustainable cities and communities are driven by urbanization and can be responded to by supply chain initiatives such as “city logistics”, “last mile deliveries” and “urban freight”.

The first two of the five goals that are considered to be directly associated with SCM rely upon process improvement and reorganisation of activities. The third goal requires decision makers to assess technological options, and make a ‘businesscase’ about investment options. The fourth goal is primarily driven by technological development and logistics (service) innovations. The fifth goal presents examples where these developments are brought into action through e.g. new ways of delivering goods to the marketplace, which may result in a new and more sustainable supply chain design. Industry-wide agendas such as “Industry 4.0” (Germany) “Produktion 2030” (Sweden) and “Smart Industry” (The Netherlands) are also driving forces of change in current supply chains. These envisage automation, digitalisation and connectivity of operations and service processes through digital technology and modular structures. Whilst competitiveness and technology are key driving forces, and result in “smart factories”, the implications for sustainability are unprecedented. Better connectivity reaches far beyond the boundaries of the individual company. For example, freight transport capacity and space at point of delivery can be shared amongst logistics service providers (horizontal collaboration), and users and consumers can become more active in such sharing (co-creation through vertical collaboration), a trend that is often referred to as the sharing economy.

196

A. Halldórsson

SCM concepts: SCM as an academic discipline and a profession drives the sustainable agenda through the development of concepts and frameworks that are taught in the class-room and brought into action through research and interaction with professionals. These concepts can be classified in two categories: (1) sustainable SCM concepts, e.g. closed-loop supply chains (retrieve rare materials from products and use as resource again, possibly after some pre-processing), recycling, reverse logistics, green purchasing and social procurement; (2) SCM concepts that have implications for sustainability such as load factors in freight transport, supplier assessment and selection, last-mile logistics and slow-steaming (reducing speed and delivery flexibility as shipments may be re-routed). In this second category we even find concepts such as just-in-time deliveries and time-based distribution, which are very popular principles in industry, but can be in sharp contrast with sustainable development. Faster and more frequent deliveries in low volumes may reduce inventory (profit) but entail increased energy consumption (planet) and a more stressful working environment with narrow and inflexible time windows of deliveres (people).

9.3 Knowing: Sustainability and SCM (Basic) The three dimensions of sustainability (TBL) are relevant to all stages of the supply chain, ranging from sourcing through manufacturing towards logistics and distribution.

9.3.1 The TripleTriple-Bottom Line Derived from the principle of the triple-bottom line (TBL), Table 9.1 provides examples of how sustainable practices in SCM have been operationalised for three different stages in the supply chain: sourcing, manufacturing and logistics. Purchasing and supply management of raw materials, components, services have traditionally strongly focused on costs as a foundation for sourcing strategies and decision criteria. To get a unit price as low as possible from suppliers while simultaneously reducing the cost of buying (costs of contacting suppliers, contracting, and control) is still a high priority, but no longer the only priority. Sourcing strategies have great responsibility in converting the sustainable vision communicated to the marketplace through brand image, advertising, and mission statements into actions through collaborative efforts with suppliers to reduce negative climate impact of e.g. freight transport. News headlines have diligently disclosed practices that have had a negative impact on companies’ image and reputation. One such example is the so-called ‘horse-meat scandal’ that emerged in Europe during the spring of 2013; suppliers (food processing and—production) and their customers (e.g. retailers) were accused of selling meat that contained undeclared horse meat as beef. To counteract this, the sourcing part of organisations have implemented a formalised approach

9 Actionable Sustainability in Supply Chains

197

Table 9.1 Sustainability at different stages in the supply chain Profit Planet Sourcing

Transport savings (Supplier evaluation and—monitoring) Improved product quality Reduce image risk Reduced total cost of ownership

Manufacturing

Increased job satisfaction and—productivity Resource utilization Energy efficiency Reduced cost of control (certification, audits, follow-ups) Higher margin for eco-friendly products

Logistics

Consolidate shipments to customers Increased resource utilisation (trucks, warehouses) Savings through increased re-use of materials and components

People

Local sourcing, reduce demand for transport Suppliers switching transport mode Sourcing strategy that consolidates shipments Supplier selection and assessment (by e.g. ISO 14001) Re-use of packaging materials Elimination of waste and excess use of resources in production process through e.g. lean production principles Modular design and use of environmentally friendly packaging Reduced energy consumption Replace hazardous materials Recycle materials (circular economy principle)

Supplier selection and—assessment Codes of conduct for safe working conditions (e.g. absence of child labour, union rights)

Design of environmentally friendly supply chain Reduce energy intensity through transport modality Slow-steaming/reduce speed Increase fill rates IT to connect customers for improved home delivers Collaboration between logistics providers

Training and education of employees in transport and warehousing (e.g. eco-driving) Warehouse layout to minimize walking distance Automation of loading/unloading Respecting driving and resting time rules Reduced congestion in urban areas

Automation of physical or monotonous work in order to improve working conditions Prevention of work accidents In-service training of employees Job rotation and enrichment Labour equity In-service training of employees 3-D printing of spares and rare items reducing inventory and transport

Note Sourcing includes purchasing and supply management (up-stream). Manufacturing encompasses all in-house operations. Logistics includes distribution, transport and warehousing (downstream)

198

A. Halldórsson

with a strong focus on social responsibility, through e.g. supplier audits and—certifications, and by use of industry-wide codes of conduct to reduce transaction cost and increase transparency to other stakeholders in the supply chain. Production has great potential in sustainable development through better use of materials and waste reduction, as well as improving working conditions and—satisfaction of employees. Finally, logistics has a great potential in improving freight transport operations and the choice of transport mode as complementary efforts to more technical solutions of e.g. increased use of renewable energy and cleaner vehicle technology. More recently, the role of “people” in the logistics system has been highlighted, both in terms of working conditions in transport and warehousing, and the impact that training for eco-driving can have on e.g. fuel consumption and carbon emissions. Hence, behaviour and managerial decisions may play a big role in operationalizing sustainable SCM, next to technological developments and governmental legislation.

9.3.2 TBL Related to SCM Practices To move beyond potential conflicts of interests and towards a win-win situation, the TBL dimensions at the various supply chain stages in Table 9.1 must be understood in relation to current SCM practices; some of these are fully in line with TBL and should be enhanced whereas others can be seen as conflicting with TBL. By combining the framework by Halldórsson et al. (2009) with Wu and Pagell (2011), the TBL initiatives can be brought into play by relating them to existing SCM practice in the organisation in at least one of the following four ways: Integrate: The particular sustainability indicator is fully consistent with SCM practice. For example, automation of loading and picking in warehouses will reduce lead times and labour costs whilst simultaneously improve health and safety conditions for workers. Align: Sustainability is complementary to SCM, but to release the full potential, trade-offs must be addressed. For example, to switch to more environmentally friendly modes of transport (e.g. from road to rail) may increase time of delivery and hence the perception that customer service is worse, and inventory costs may rise. Customers and suppliers must engage in a dialogue to align strategic priorities such as time and costs. Re-conceptualise: When alignment of strategic priorities and processes is not sufficient to improve sustainable performance, companies may need to re-conceptualise their system and introduce structural changes to enhance sustainable performance. An example of this is the so-called “slow steaming” concept, borrowed from the maritime sector, but also used to explain how the down-stream part of the supply chain, closer to retailing and consumption, can benefit from reduced speed (lower fuel consumption, allow for more use of sustainable modes of transport) across different actors in the chain. Replace: The most radical but also most difficult to implement is replacement of current business logics and-practices. Here, companies may need to question the logic

9 Actionable Sustainability in Supply Chains

199

of the entire supply chain from e.g. building upon responsiveness and time-based competitions to take a firmer eco-centric approach. This might require innovation of business models and logistics services, in which e.g. parts of the logistics capacity are regarded as an “open source”, i.e. readily available to use by a much larger group of stakeholders than today.

9.4 Doing: Building Blocks of Actionable Sustainability (Advanced) The scope in Fig. 9.1 and initiatives in Table 9.1 represent the “knowing” of sustainable SCM. A next step is “doing”; to mobilise actions that create impact along the horizontal and vertical dimensions in the framework. Actionable sustainability helps SCM professionals to conceptualise options in a fashion that can be acted upon. The following three building blocks help us determining the scope of actions: (i) Cause and impact: Whilst sustainability often implies an internal perspective of the company’s operations, it is important to understand that, from a supply chain perspective, the root cause of sustainability issues ought not to be separated from its impact. Stressful working conditions, a social dimension of sustainability, in a warehouse of a logistics service provider may be caused by a responsive supply chain design of the customer, that in turn has been structured to meet strict demand of consumers of almost instant homedeliveries. For example, a “next day delivery” option in e-commerce that can be selected at customer check-out may lead to pressure on individuals and irresponsible behaviour in e.g. freight transport (speed driving, fuel consumption) or warehousing (damaged goods). To avoid scattered responsibilities and interests among individual actors, working with sustainability in SCM requires joint-up thinking, from the point of use or consumption and all the way back to the upper-tier suppliers. The scope of managerial responsibility should be defined accordingly. (ii) Relationships in triads: A triad (a set of three actors) is a useful metaphore to explain the interaction between the common actors in the supply chain: logistics provider, customer/shipper and consumer. These interact in a “co-creation” of not only services but also outcome that can be defined in terms of the triplebottom line (TBL). Such an interactive approach and focus on relationships between actors as the foundation of success is very much in line with the literature on e.g. strategy management and industrial marketing. Companies that truly want to mobilise sustainable actions must not only work on process improvement within the boundaries of organisations. Rather, foundation of success and business strategies that create value is also based upon how well an organisation manages relationships with customers and suppliers (see e.g. Kay 1993).

200

A. Halldórsson

(iii) System levels (services and processes): Companies must move across system levels in their improvement efforts. Load factors, for example, are a common measure of capacity utilisation in freight transport. A high load factor has been positively associated with environmental sustainability, i.e. a full truck load is better than a half filled truck (relatively small freight). In this case, the achievement is measured around the system boundaries of the vehicle (truck) that is operated by logistics services providers. However, if we extend the system boundaries to also include the customers of LSP’s, and even the consumers/users that receive the shipments, too much emphasis on high load factors alone as an ultimate performance measure can backfire in terms of sustainability. For example, in the case where up to 5–10% of products shipped to the marketplace (e.g. spare parts from OEM manufacturer to car dealers) are returned to the OEM (distribution centres) or even up-stream suppliers, the original load factor criterion can be misleading. Another example is food products; when pushed aggressively to the marketplace but not sold before their expiry date these immediately turn into “waste”; both in terms of products that are thrown away, and also irresponsible use of natural resources in growing food products (people, planet), and of logistics resources in bringing these to the marketplace (profit). Sadly, much of the food wasted in particular parts of supply chains could in principle be used to address the hunger problem in other regions of the world.

9.5 Mobilising Actions Through Modes of Innovations (State of the Art) Implementing sustainability in global supply chains must translate potential identified in the scoping phase into actions. As mentioned earlier, such actions can be mobilised through current practices by integration, alignment and reconceptualisation, and ultimately, replacement of current practice. In particular the latter two require a more radical approach than incremental efforts allow for. It is therefore useful to borrow from the logic of innovation to explain changes in the different characteristics of logistics services as well as the structure of the supply chain that may be needed. Such an approach is also motivated by customers of logistics service providers who increasingly see innovation as a ‘given skill’ when buying logistics solutions. Approaching this task as a disruptive innovation rather than a mere incremental improvement of current practices indicates a strong desire to change, and helps the company to develop an ability to address the challenge raised by TBL, to accommodate new and emerging customer needs, and to release the potential of new technology. By using the vocabulary of Gallouj and Weinstein (1997), Table 9.2 identifies four modes of innovation that focus in particular on logistics services (the third component in Table 9.1).

9 Actionable Sustainability in Supply Chains

201

Table 9.2 Mobilising sustainable SCM through logistics service innovations Mode of innovation Example Ad hoc Reacting to a problem posed by a given client. Original solutions used to react to new problems, i.e. used under different circumstances. Often jointly (i.e. service provider and customer) produced

Logistics service providers use current services to deliver into the customer’s new fulfilment modes. Use of same capacity across a variety of fulfilment modes, to assist customers in moving from separated multi-channels towards integrated omni-channels. This can reduce the energy footprint of transport, and improve the working conditions for truck drivers (working hours, waiting times)

Formalisation Focus on visibility and bringing service characteristics “in order” through standardization and modularization

Reduced speed as a consequence of “slow-steaming” can be counteracted by hyperconnectivity of the Physical Internet (see Chapter 31). This requires standardised interfaces among actors in the supply chain, both technical and organisational. Further, the use of modularised packaging materials and containers, makes materials handling, transport, and delivery more efficient across all three dimensions of TBL Automotive manufacturers have become part of last-mile fulfilment solutions by offering “in-car delivery” access to the vehicle. Through an app on their mobile phone, consumers can authorise a one-time access to logistics providers to have their goods delivered into their own car. With the private car as delivery location, logistics providers are more likely to successfully deliver a package rather than repeatedly attempting to deliver to customers home address where no one is at home to sign for the shipment. This can reduce the number of trips in residential areas

Recombinative Also called architectural innovation through a new combination of various characteristics of existing services

Radical A new service is created, based upon a new system. No commonalities with old service, new competences needed

TBL achieved by an open-network to logistics services. Principles and technology of the Physical Internet used to redefine the global supply chain by enabling universal hyperconnectivity of logistics services. A new business model allows for new ways of horizontal as well as vertical sharing of logistics resources in the supply chain. True co-creation of TBL in logistics service triads

202

A. Halldórsson

The examples in Table 9.2 indicate that further development of sustainable SCM can be mobilised through various modes of innovation, which will not only redefine or reconceptualise, but that require changes in structures and services characteristics, and that also bring new players into the service triad. Behind this lies the sense of urgency that motivates multiple stakeholders in the supply chain as well as the emergence of sustainable SCM as a concept. Opportunities for improvement exist both horizontally and vertically in the supply chain, and a “true sustainability” should refer to both dimensions, and within these, sustainability should be operationalised through the principle of the TBL. To mobilise options into actions, companies must seek to relate the indicators of TBL to current practices and/or refer to various modes of innovation of e.g. structures and services so these principles can be acted upon; managers as well as academics must strive for an actionable sustainability.

9.6 Further Reading Seuring and Müller (2008) and Carter and Rogers (2008) provide an early but much discussed account of sustainability in SCM. Kovacs (2008) adds to this an important social responsibility perspective. Later, Gimenez et al. (2012) broaden the perspective on sustainability by relating TBL to SCM (e.g. the vertical dimension in Fig. 9.1). Gualandris et al. (2015) explain how companies can address for accountability for their sustainability behavior in the wider supply chain (the horizontal dimension in Fig. 9.1). Another extension of the responsibility of SCM is offered by Genovese et al. (2016). Here, sustainability in SCM is related to circular economy and resource utilisation. Zhu et al. (2008) and Mejías et al. (2016) emphasie application by their perspectives on implementation of sustainability practices. Finally, Montabon et al. (2016) put sustainability highest on the agenda. To make supply chains truly sustainable they must build upon an ecological-dominant logic, where the interests of people and planet given higher priority than profit. Acknowledgements The support of the Swedish Energy Agency for funding the projects “The Fifth Fuel” and “Energy Efficiency of Logistics Services” is gratefully acknowledged.

References Carter CR, Rogers DS (2008) A framework of sustainable chain management: moving toward new theory. Int J Phys Distrib Logistic Manag 38(5):360–387 Gallouj F, Weinstein O (1997) Innovation in services. Res Policy 26(4–5):537–556 Genovese A, Acquaye AA, Figueroa A, Koh SCL (2016) Sustainable supply chain management and the transition towards a circular economy: evidence and some applications. Omega 66:344–357 Gimenez C, Sierra V, Rodon J (2012) Sustainable operations: their impact on the triple bottom line. Int J Prod Econom 140:149–159

9 Actionable Sustainability in Supply Chains

203

Gualandris J, Klassen RD, Vachon S, Kalchschmidt M (2015) Sustainable evaluation and verification in supply chains: Aligning and leveraging accountability to stakeholders. J Operat Manag 38:1–13 Halldórsson Á, Kotzab H, Skojett-Larsen T (2009) Supply chain management on the crossroad to sustainability: a blessing or a curse? Logist Res 1:83–94 Kay J (1993) Foundations of corporate success. Oxford Kovács G (2008) Corporate environmental responsibility in the supply chain. J Clean Prod 16:1571–1578 Lambert DM, Cooper MC, Pagh JD (1998) Supply chain management: implementation issues and research opportunities. Int J Logist Manag 9(2):1–19 Mejías AM, Paz E., Pardo JE (2016) Efficiency and sustainability through the best practices in the logistics social responsibility framework. Int J Operat Product Manag 36(2), 164–199 McKinnon AC (2016) Freight transport deceleration: its possible contribution to the decarbonisation of logistics. Transp Rev 36(4):418–436 Montabon F, Pagell M, Wu Z (2016) Making sustainability sustainable. J Supply Chain Manag 52(2):11–27 Sarkis J (2003) A strategic decision framework for green supply chain management. J Clean Prod 11:397–409 Seuring S, Müller M (2008) From a literature review to a conceptual framework for sustainable supply chain management. J Clean Product 16(15):1699–1710 Stock R (1992) Reverse logistics. Oak Brook, IL, Council of Logistics Management, p 214 Wu Z, Pagell M (2011) Balancing priorities: decision-making in sustainable supply chain management. J Operat Manag 29(6):577–590 Zhu Q, Sarkis J, Lai K-h (2008) Confirmation of a measurement model for green supply chain management practices implementation. Int J Prod Econ 111:261–273

Chapter 10

Human Resource and Knowledge Management Matthias Klumpp, Sascha Bioly and Thomas Neukirchen

Abstract Future innovations in logistics and supply chain management are driven by technological, regional, economic and social changes as well as sustainability and resource restrictions. They will not occur without sufficient attention for the human contribution, in particular the qualification, competence and motivation of the workforce in the logistics domain. Worldwide, about 50 million people—or the entire population of South Korea—are employed in the SCM, logistics and transportation sectors. Consequently, human resource and knowledge management plays an important role, more so when considering demographic developments (ageing, migration) and increasing globalization. Whereas in the past, many blue-collar jobs in logistics such as truck driving merely required a basic school education and rudimentary qualification levels, these jobs now require increased competences due to improved technology interaction, e.g., barcode and RFID systems, fleet management or toll and truck steering concepts and finally artificial intelligence applications. The same is true for many white-collar jobs in logistics, exemplified by the increasing number of university graduates employed in the sector. This is especially true for specific fields such as logistics information technology, contract logistics and supply chain innovation and design, which in turn leads to the question of how to assess qualitative and quantitative competence levels in the SCM and logistics sector. Such an analysis example is provided here with the Berufswertigkeit concept (reference framework for competence levels). This may lead to a new paradigm in HR and knowledge management for SCM and logistics: whereas past education was mainly driven by formal qualifications and therefore personnel groups (white-collar/blue-collar), future conM. Klumpp (B) FOM University of Applied Sciences and University of Duisburg-Essen, Essen, Nordrhein-Westfalen, Germany e-mail: [email protected] S. Bioly Forschungszentrum Jülich, Projektträger ETN, Jülich, Germany e-mail: [email protected] T. Neukirchen FOM University of Applied Sciences, Essen, Nordrhein-Westfalen, Germany e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_10

205

206

M. Klumpp et al.

cepts may focus on an individual analysis of gaps and potentials based on elaborate evaluations. In addition, modern concepts like edugaming are outlined as examples for future qualification and training concepts for logistics personnel.

10.1 Introduction: Logistics Personnel and Qualification (Basic) The supply chain, logistics and transportation industry has become a knowledgeintensive service industry, similar to other sectors such as the banking, insurance, engineering or the healthcare industries (Klumpp 2016). In the past, mainly manual labor and only basic (mostly vocational) qualifications were commonplace, both in blue-collar (e.g., truck drivers, warehouse personnel) as well as white-collar (e.g., dispatchers, managers) employment in logistics. Today, qualification requirements are different und significantly higher and they are crucial for business success in terms of being a strategic success factor (Rao et al. 1998; Lancioni et al. 2001; Mangan and Christopher 2005; Wu 2007; Esper et al. 2007, Rogers and Braziotis 2016). Such rising qualification requirements are a consequence of technological innovations such as barcoding, RFID, GPS, and automated systems in physical logistics processes. At the same time, the overall complexity of logistics processes has risen manifold, thereby making the tasks of the planning, management and control of the logistics function even more demanding. This is because supply chains became global, with a multitude of actors and stakeholders as well as increasing demands of customers regarding quality, sustainability, price and cost-efficiency, next to flexibility and speed in delivery (Klumpp et al. 2013; Rogers and Braziotis 2016). Therefore, competence requirements for logistics personnel have grown due to the multitude of complex tasks in logistics management—to the point where innovation and growth may be hampered due to missing competences and regulations regarding process standards (Zijm and Klumpp 2016, 14–19). In order to provide a quantitative overview of these developments some statistical data may serve as a starting point for the discussion on logistic qualification. In 2015, 2.9 million people were employed in the logistic industry for example in Germany. Comparing this with the overall GDP in Germany (3.356 trillion US $ in 2015 according to the World Bank) yields a rate of 1,157,162 US $ of GDP per logistics employee. Extrapolation of this rate to the world’s GDP leads to a global estimate of logistics employees of 63.5 million persons (high estimate). Providing a low estimate with only two thirds of the GDP per logistics employee yields an estimate of 42.3 million persons in the sector. In this chapter, we therefore take the average of about 50 million persons employed in the SCM and logistics sector as a starting point for a global estimate for all personnel employed in the logistics sector. It is obvious that a clear-cut figure is impossible to retrieve mainly due to data gathering and definition problems (e.g. determining whether people in industrial companies, for example in the automotive sector, work with logistics tasks; we are familiar with data only for Germany) (Table 10.1).

10 Human Resource and Knowledge Management

207

Table 10.1 Global personnel estimate in the SCM and logistics sector GDP 2015 US $ Germany World

3,355,772,000,000 73,502,341,000,000

Logistics Low estimate employees 2015

High estimate

2,900,000







42,346,299

63,519,449

Sources World Bank (GDP), BVL Germany (Logistics Employees)

Many arguments could be presented in favor of or against the defined estimate range, mainly questioning the rate of logistics employees per GDP for Germany compared to other countries. Such arguments may address the question of population and population density (high for Germany), infrastructure (well developed in Germany), logistics professionalism or technology (also very high in Germany as according to the World Bank Logistics Performance Index, see http://lpi.worldbank.org, Fig. 10.1). Most arguments would have impacts in the direction of larger estimate numbers as they might allow only for a lower GDP per personnel than Germany; e.g., with lower infrastructure standards as well as technology implementation one would need more personnel for identical transportation and logistics tasks compared to Germany. This leads to the assumption that the presented estimate number is mainly on the lower safe side and the real number of logistics employees worldwide may well be larger. The logistics industry has experienced a sector growth of about 1–2% above the overall average economic growth per annum. Combined with the earlier mentioned technological progress, there are sufficient arguments to pay special attention to education in the logistics industry: firms need qualified personnel in the logistics field not only because of the described technology and supply chain organizational changes; but also to cope with the above-average growth while keeping the transport chains cost-efficient as well as sustainable for the customers and society at large (Klumpp 2016; Rogers and Braziotis 2016). In order to implement this, qualified personnel is needed, in some cases very scarce which in turn may be used as an indication where actions from HR and training as well as knowledge management is required. In Fig. 10.2, the most important qualification fields with “missing personnel” are reported in a recent logistics survey from Germany. Obviously, especially specific academic qualifications in the fields of computer science (IT), management and engineering are in need, matching the “cross-discipline” nature of logistics and supply chain management. This chapter is structured as follows: Sects. 10.2 and 10.3 further outline the basic concepts, terminology and requirements regarding qualification and human resource management in logistics. On an advanced level, Sect. 10.4 describes the tools and concepts for HR management—whereas Sect. 10.5 provides such tools and concepts for knowledge management in logistics. Addressing a state-of -the-art level, Sect. 10.6 is outlining the innovative Berufswertigkeit evaluation concept for competences in the logistics field; Sect. 10.7 is further describing a new research and development approach with edugaming for training and knowledge management.

208

Fig. 10.1 Logistics performance index—leadership 2016 (World Bank)

M. Klumpp et al.

10 Human Resource and Knowledge Management

209

Fig. 10.2 Personnel and qualification requirements in logistics (BVL Germany)

10.2 Case Study: Bohnen Logistics Bohnen Logistics was founded in Germany already in 1926 when Johann Bohnen started driving individual transport shipments for the local brick industry near Düsseldorf, Germany. At this time still with horses and carriage, he provided the missing transport link within the supply chain of core building corporations. Soon he realized that innovation and technology is the key to business success, especially in transportation and logistics: He was one of the first entrepreneurs to buy and use road trucks instead of horses. This enabled the business to grow in terms of employees as well as transport distance—soon he was doing business within a large part of western Germany along the river Rhine. When his son Reiner Bohnen took over in the 1950s, the major challenge was to enlarge the service range and the geographical reach at the same time, going international within Europe. New customers and value-added services like warehousing and IT support systems were acquired and several other company locations established. Throughout the corporate history of success and growth, also the third generation CEO, Jürgen Bohnen, emphasizes the imminent importance of a “human workforce and innovation culture”: Based on a very engaged, emotional but also competitive and demanding cooperation with all employees, Bohnen Logistics is convinced to provide the necessary basis for new services, technology innovation and therefore further growth.

210

M. Klumpp et al.

Today, Bohnen is a valuable service brand in Germany, standing for innovation, flexibility and high-quality service and working with customers like Danone and CWS Boco. The company received the NRW.Invest Award for a top-notch new warehouse with automated services and a size of 40.000 m2 in Duisburg, the largest inland port of Europe. The sights of the company and the employees are still onto new innovations, further automation as well as added services for customers. The bottom line of this successful corporate story within the logistics industry is the fact that qualification and competences of employees go hand in hand with innovation, including technology as well as business case innovation and services offered. The need for innovative training and qualification concepts finds support by results of an internal survey conducted by Bohnen Logistics amongst employees. Concerning the blue collar-workforce, employees were asked about the role of logistics training. Regarding self-assessment by the interviewed, the importance of ongoing qualification was stressed with respect to changes in regulations and law. Willingness to pursue work-related training measures was (i) given in all but those considering retirement, (ii) conditional on training happening during work hours and resulting in visible and material (monetary, status) rewards, and (iii) conditional on student-centered (as opposed to teacher-centered) learning, engagement, fun, intrinsic motivation. Rising complexity and changes in preferences concerning ongoing qualification and training, particularly student-centeredness, intrinsic motivation and fun, and visibility of advances made can be dealt with using modern e-learning and edugaming approaches such as those outlined in Sect. 10.7 of this chapter. It is important to recognize that qualification and training in logistics are not a “nicety” to employees but a “necessity” for strategic management in order to enable innovation and business as well as earnings growth. (Source: Personal communications and www.bohnen-logistik.de)

10.3 Terminology and Competence Requirements Definitions regarding qualification and competence measurement as well as continuing education and training are often fragmented or imprecise. Competence is defined as “the ability to successfully meet complex demands in a particular context. Its manifestation, competent performance (which one may equate to effective action), depends on the mobilization of knowledge, cognitive and practical skills, as well as social and behavioral components such as attitudes, emotions, values and motivations” (Hakkarainen et al. 2004). Competence demonstrates also the level of student achievement in a science education context; competence is there-

10 Human Resource and Knowledge Management

211

fore not only skill, qualification or knowledge, but all these factors together constitute the basis for the competence of an individual person (Liu 2009; Klumpp 2016). The term continuing education is also used as for example further education or training—all dedicated to the same question of on-the-job or parallel-to-the-job qualification (Cervero 1988; Mezirow 1991; Jarvis 1995; Hanft and Knust 2009). Already in 1970, the German Education Council determined continuing education in the German education structure: it can be defined as continuation or resumption of learning after a first degree; continuing education therefore usually might begin after entering the workforce (Bildungsrat 1970). Furthermore, continuing education includes formal, informal and non-formal learning (Marsick and Watkins 2001; Hofstein and Rosenfeld 2008; Dabbagh and Kitsantas 2012; Klumpp 2016). Formal learning means a regulated and structured continuing training, which is organized by institutions and where students have the chance to gain acknowledged degrees and certificates. Informal learning indicates continuing education in project groups, networks and coaching without acknowledged degrees or certificates—but with a recognizable teaching-learning setup, whether be it personal or also electronically or virtually. Non-formal learning is learning by doing or learning on the job without even standardized or organized learning environments and processes. Further, this can be distinguished into general (not practice- and profession-oriented), vocational (practice- and profession-oriented by deepening practical experience) and higher (education at research universities and universities of applied sciences). Generally speaking, continuing education may provide advantages for all stakeholders. These advantages are defined in economic and social dimensions, with three levels each (European Centre for the Development of Vocational Training 2011): • Macro Level: Advantages for a whole society – Economic profit: Economic growth and labor-market outcomes – Social profit: Crime reduction, social cohesion, health and intergenerational benefits • Meso Level: Advantages for enterprises and groups – Economic profit: Firms performance and employees productivity – Social profit: Inclusion disadvantaged groups • Micro Level: Advantages for individuals – Economic profit: Employment opportunities, earning and career development – Social profit: Life satisfaction and individual motivation The logistics sector is also characterized by many people changing career tracks and industries, even in mid-career. Also for “newcomers” there are ample possibilities within continuing education facilitating the acquisition of specific logistics skills (for an example overview of professional tasks and trainings in Germany see: Berufswelt Logistik 2017). To continue education with many years of business practice experience (minimum one year) in the logistic sector there are two different ways to

212

M. Klumpp et al.

Table 10.2 Differences between e-learning 1.0 and e-learning 2.0 E-learning 1.0 2.0 Learning environment

A closed area in the internet supporting content and tools

An open platform to the internet supporting tools for generating content

Teachers

Transfer all known resources into this closed area Consume the given content

Define boundaries and offer resources Configure their personal learning environment (PLE) to generate own content

Students

extend individual competences in Germany: through academic continuing education programs at universities (part-time or full-time) that offer an academic degree, or through practice-oriented continuing professional education (Klumpp 2016). In order to support success in continuing education and to motivate the current generation, suitable learning tools such as e-learning platforms developed within the 21st century. Because information and communication technologies such as smartphones and notebooks find their way into everyday life, e-learning is an important pathway, especially for lifelong learning scenarios. The main advantage of e-learning is the possibility of receiving information anytime and anywhere. Besides e-learning scenarios without physical presence in a classroom, blended learning concepts have been realized in which traditional face-to-face learning situations are combined with e-learning elements (Klumpp 2016). Most educational institutions offer their students e-learning platforms to support the face-to-face learning sessions with additional information. Two main software concepts are Moodle, which was developed at the University of Cologne, and Blackboard, developed by Blackboard Inc., Washington D.C. According to a definition of Web 2.0 in 2005 e-learning technologies were also developed to stimulate an active participation of the learner. Social software has been conveyed into learning environments, like wikis, podcasts or blogs. The boundary between teacher and learner disappears and collaborative learning scenarios gain in importance so that new technologies evolved (Blees and Rittberger 2009). Table 10.2 displays an overview of the differences between e-learning 1.0 and 2.0. Competence is a major asset when measuring logistics industries’ competitiveness—as also demonstrated by the Logistics Performance Index (LPI) published by the Worldbank, where it is one out of six indicators regarding a country-specific competence level in logistics and transportation (Worldbank 2017). Especially from the perspective of logistics practice, there are several major trends to be recognized and reviewed for future management concepts in relationship with competence and knowledge management (Zijm and Klumpp 2016): • Globalization: Open Supply Chains and scattered production plants demand for reliable logistic processes.

10 Human Resource and Knowledge Management

213

• Digitalization: High integration of information systems such as telematics, mobile handhelds, tracking & tracing, etc. • Knowledge management: The success of logistic service providers is often intimately connected with the employees’ knowledge. • Volatility of economic development: The logistics industry experiences a more severe impact of economic fluctuations than other industry sectors. • Security: Examples are attacks by pirates on ship as well as disruptions due to civil wars and natural disasters. • Sensitivity in ecological and sustainability questions: Carbon and other emissions as well as energy and resource consumption through logistics. In addition to the main subjects identified above, most high-wage countries (i.e., in Europe and America) are affected by the demographic change in their population. Especially the baby-boomer generation with age cohorts from the 1950s and 1960s implicate that after 2020 many employees will withdraw from economic activity because of reaching their retirement age. When considering the identified trends and the development of the aging structure in logistics it is obvious that innovative logistics learning solutions have to be designed by offering employees possibilities for lifelong learning. The number of logistics education, qualification and training programs increased in the last 20 years: today, in many European countries universities provide specific logistics programs as well as economic or technical programs with significant logistics content (Hildebrand and Roth 2008). However, the main challenge is to provide learning possibilities for employees without leaving their job. Employees have to increase their knowledge to tackle the tasks of logistics management in a global high-velocity supply chain environment. Therefore, flexible e-learning scenarios offer the possibility of knowledge acquisition on the job and account for the above-mentioned dependency on current trends. The integration of technical solutions underlines the employees’ capability to acquire knowledge within an e-learning scenario. The capability of logistics learning mechanisms depends on four components: temporal components, cultural components, structural components and relational components. The consideration of these four components is a major requirement for a successful learning process. The cultural component can be seen as a basis of learning because the entire logistics sector and supply chains are internationally oriented. The structural component regards the specifications of the employee’s organization to realize learning activities on-the-job: flexible in time and position. Relational components assist the collaboration and communication within a strongly cross-linked company structure and the temporal component supports the velocity of changes within the logistics sector and synchronizes them with the learning process (Esper et al. 2007). The goal of learning is to match an employee’s knowledge with the needs of the logistics industry. The needs can be displayed in the shape of a “T” (Fig. 10.3). The horizontal level displays specific logistics know-how and the vertical level displays the understanding of other company departments with connection to logis-

214

Specific logistics know-how

Understanding of departments with connection to logistics

Fig. 10.3 Employee capabilities for logistics (Mangan and Christopher 2005)

M. Klumpp et al.

tics, for example process management, engineering (R&D), production, sales or accounting departments. In a best-case scenario, operations personnel with practical experience has been equipped with management tools and competences and therefore develop into logistics managers of the future. For designing educational courses in logistics for persons already employed, the content must be carefully designed and developed. In addition, structural requirements gain in importance because of the development of the sector. These aspects are relevant because the logistics industry shows several specific characteristics: high speed and flexibility of services, significant shares of small and medium-sized companies and a typically high level of personal tacit knowledge leads to the structural requirements listed in Fig. 10.4.

10.4 HRM Tools in Logistics (Advanced) Innovations in logistics are often only associated with technical solutions. This restriction has implications for an area in which new concepts and solutions can raise great potential for optimization—the field of human resources. It is not a matter of looking at the employees as a simple production factor, but rather the organizational design of the working conditions and the work environment and thus new possibilities for the development of self-initiative and creativity to increase competitiveness. In order to achieve this, the general HR objectives and tasks of recruiting, assessment, performance review, training as well as long-term career management and

10 Human Resource and Knowledge Management

215

Scheduled for working persons Applicationoriented / practically relevant

Company independent Structural requirements of learning in logistics

Easy to understand: descriptive

Targeted

Supporting communities

Fig. 10.4 Structural education requirements for the logistics industry

development have to be applied in general as well as specifically towards logistics employees and tasks. Additionally, HR departments increasingly also have to focus on strategic, valueadding tasks, particularly in the SME sector. This requires appropriate HR tools. In the personnel departments of many small and medium-sized companies, Excel tables are still used to perform daily routine tasks, for example, on payroll accounting, holiday and personnel deployment planning. Due to the inefficient implementation of such administrative activities, there is often no time for strategic, value-adding initiatives. This is criticized by experts from Hamburg’s consulting firm Steria Mummert as part of a comprehensive 2011 study (see https://goo.gl/LCjCGl). It concludes that SMEs neglect human resources management. Anyone who performs routine tasks in human resources efficiently can focus on more important activities, such as optimizing employee productivity, increasing employee satisfaction, or professionalizing processes in areas such as recruiting, competence and talent management—which are becoming more and more important in times of a “war for talent” (Michaels et al. 2001).

216

M. Klumpp et al.

This applies to almost all industries. For logistics, some facts are aggravated though: • Very small companies: operational logistics—in the sense of physical goods handling—is dealt with by a large number of very small companies. These tend to have only little knowledge and even fewer resources for HR topics (in most cases there is no specific HR employee/management, but the company CEO (often the owner) has to fulfill HR task “on the fly”). • Small and medium sized companies (SME): There are additional challenges in the area of logistics services and value-added services. This area of the industry is highly competitive. With low profit margins, only limited funds are invested in medium- and long-term planning. • Large companies: This field shows a fundamental challenge for the industry, i.e. the general industry and employer attractiveness. For example, Amazon is certainly not (yet) a typical logistics company, but develops in some aspects in this direction. The company is setting up its own hubs1 or plane and vehicle fleets, and develops alternative delivery systems (delivery by drones). The example of the personnel payment shows at this point only the problem per se. In the distribution center in Bad Hersfeld in the center of Germany, the company pays a warehouse employee 10.01 e per hour in the first, 11.59 e in the second and 11.71 e in the third year. According to the wage agreement for shipping, it would have to be 11.77 e in the first three years, and 10.93 e in the logistics sector. In Leipzig, Eastern Germany, the hourly rates are 9.55/10.47/10.99 e in the first three years. Per wage agreement, it would have to be 11.39 e in all years, and in the logistics sector 9.17 e, in the second and third year 9.61 e.2 So, in general, payment levels are not very attractive, at least in the blue-collar section of transportation and logistics. This is why companies try to be attractive for employees with other measures such as training support, bonus payments, company equipment or health support as shown by the following example from a German logistics HR survey (Fig. 10.5). Personnel development is one of the key pillars in the area of company and organizational development. The effective training, coaching and further training of the employees and managers in the company combined with a goal-oriented team development is one of the most difficult tasks to be accomplished by today’s Human Resources Manager. This is exemplified by Fig. 10.6, addressing the use of personnel development tools in logistics. Some instruments—at least for the case of Germany—are obviously in place such as performance review and training measures. However, others like e.g. assignments abroad or long-term career support (“high potentials” etc.) are still seldom found in the logistics sector though advancing in all other industries.

1 Cf.

https://www.bloomberg.com/news/articles/2017-01-31/amazon-plans-new-air-hub-in-kentuc ky-to-support-fast-deliveries. 2 See https://goo.gl/jh8EHY.

10 Human Resource and Knowledge Management

217

Fig. 10.5 Use of HR tools in logistics (BVL Germany)

Fig. 10.6 Use of personnel development tools in logistics (BVL Germany)

This represents an often-found “gap” in HR and personnel development, causing the transportation and logistics sector to be recognized as a “latecomer” to new developments in HR management.

218

M. Klumpp et al.

10.5 Knowledge Management Tools in Logistics Knowledge management is a comprehensive term for strategic and operational activities that deal with the handling of knowledge. Contributions to this topic are often interdisciplinary, including business economics, social sciences, and computer science. As a result of today’s knowledge- and innovation-oriented communication age, the knowledge capital available in the company is increasingly becoming the decisive factor in production, which is taking shape as factor of production alongside labor, land and capital (Abdih and Joutz 2005; Probst et al. 2006). Often, information is not where you need it. Companies have recognized that efficient management as well as dealing with corporate knowledge help to increase their own competitiveness. Critical to the approach of knowledge management from a scientific point of view is, above all, an undifferentiated concept of knowledge, which is often not adequately delimited by the terms “data” and “information” (Meyer and Sugiyama 2007). In addition, knowledge can be subdivided into different categories, each of which has to be treated differently: • Explicit knowledge is formulated and reproducible knowledge. It can be mediated without difficulty by a formal, systematic language, such as words and numbers. It can be logically reproduced and described and therefore it represents specific or methodological knowledge. • Implicit knowledge, on the other hand, has a personal quality that makes it difficult to formalize and communicate. It is hidden knowledge that cannot be articulated. Moreover, it is strongly motivated by the related actions, obligations and co-operation within a specific context. Polanyi, in his theory of implicit knowledge, explains human knowledge with the proposition “that we know more than we know” (Polanyi 1985). If this distinction is applied to the personal level of the competence and knowledge of one person, two specific forms can be derived: • Individual, explicit knowledge is labelled as embrained knowledge. It is a conscious knowledge that depends on one’s own conceptual abilities and can be activated consciously, e.g. specific knowledge. This knowledge can be transferred by rules, instructions or information and communication technologies. • Individual, implicit knowledge is referred to as embodied knowledge. It is an action-oriented knowledge and results from the experiences already gained. This includes cognitive abilities such as concepts and experiences, but also skills such as the fine motor skills of a dentist or the ability to dance on a rope. The transfer of this knowledge requires intensive interaction processes and cannot be ordered by directives or controlled by the price mechanism. However, the sum of the explicit and implicit knowledge that the individual members of the organization possesses is not yet an organizational knowledge per se. Organizational knowledge arises only from the coordinated collaboration of the organizational members. The incorporation of individual knowledge and knowledge into

10 Human Resource and Knowledge Management Table 10.3 Knowledge dimensions (Lam 2000, 491ff.) Individual knowledge

219

Collective knowledge

Explicit knowledge

Embrained knowledge Conscious, verbal skills and competences

Encoded knowledge Knowledge depicted in rules and procedures

Implicit knowledge

Embodied knowledge Internalized ability

Embedded knowledge Organized by organizational routines and mental models

specific “organizational settings” is the prerequisite for developing organizational knowledge from the knowledge of individual organizational members (e.g., Hecker 2012; Nonaka 1994). This collective knowledge can also be explicit or implicit: • Explicit, collective knowledge is called encoded knowledge. This knowledge exists in companies in the form of rules and procedural guidelines that are applied in a company. They are expressed, for example, in organizational models, organizational charts, management principles, or strategic concepts pursued by the company. This knowledge can be documented. • Implicit, collective knowledge is called embedded knowledge. It is found in companies mainly in the form of organizational routines as well as “mental models” shared by the organizational members. This means the implicitly used everyday theories of action by the members of the organization. Taking these four categories together, we arrive at a two-by-two matrix for the question of individual versus collective knowledge in the explicit or implicit form (see Table 10.3). Nowadays implicit knowledge in companies is particularly important as a source of sustainable competitive advantages (e.g. Eisenhardt and Santos 2002). It is particularly difficult to imitate if this knowledge can be anchored organizationally in knowledge management processes. It is not enough to accumulate and store much information or to employ employees with specialist knowledge. Individual, implicit knowledge is the basis for knowledge management, but does not in itself represent a sustainable competitive advantage for companies, because individual knowledge providers can be lured away. In this case, they leave much of their explicit knowledge in the form of records. However, their implicit individual knowledge is lost to the company. Organizational scientists Nonaka and Takeuchi developed the most familiar model of knowledge management in 1995 with the so-called “knowledge spiral”. The core is that the continuous exchange between explicit and implicit knowledge is the prerequisite for the generation and transfer of organizational knowledge. In this way, implicit knowledge can be spread throughout the organization and can be constantly enriched at the same time. In order for organizational knowledge to be created, the individual implicit knowledge of the organizational members has to go through a dynamic transfer process. For this purpose, explicit and implicit knowledge are com-

220

M. Klumpp et al.

bined into four different forms of knowledge transfer: socialization, externalization, combination and internalization (Nonaka and Takeuchi 1995). Socialization transmits knowledge “from implicit to implicit”, that is, largely without language. Instead, “learning by doing” via observation, imitation and exercise is central. Thus, children learn to cycle by practicing pedaling, steering and balance until they can. A typical example of socialization in day-to-day operations is the integration of a new team member into the group’s thinking and action routines. Externalization turns implicit knowledge into explicit. However, this conversion is always only partially possible. Prerequisite for the externalization of implicit knowledge is intensive personal communication, e.g. in quality circles or interdisciplinary teams. Using analogies and metaphors, the participants try to make their implicit experience knowledge accessible to each other. The combination consolidates different explicit knowledge. Since the combination of knowledge is not linked to “face-to-face” contacts, it can be supported by information technology. Conventional information technologies deal exclusively with this form of knowledge transfer. They thus take into account only a small part of the relevant knowledge. With internalization, explicit knowledge is (partially) again transformed into implicit knowledge, but in an enriched, more complex form. This is done by learning individuals or groups of action routines that were previously explicitly formulated. The safe control of routines allows complex activities to be carried out “as in sleep”. They only require a reduced attention. From a technical point of view, the use of a knowledge management system can be useful. The basic functions of such systems are: • Content Management—the storage of explicit knowledge in documents, database entries, images, as a platform for the exchange of knowledge, often as an intranet form. • Information retrieval—Find and retrieve the needed information. • Visualization and Aggregation—The structuring of knowledge and representation of not easily explicable knowledge (graphics, pictures, drawings). • Collaboration—Collaboration of people and groups for example with groupware. Especially for small and medium-sized enterprises, it is a tough challenge to introduce knowledge management alongside day-to-day business. Because sustainable “deep” change requires time and effort. Knowledge management is not fundamentally new, but the use of information technology and the application of innovative organizational methods can make the handling of knowledge in the company more systematic. In recent decades, the term knowledge society has become increasingly important. This term was used in 1966 by the sociologist Robert E. Lane. In 1973, Daniel Bell developed the concept further with his study “The Coming of Post-Industrial Society” (Bell 1973). While labor, raw materials and capital have a central role in industrialized societies; theoretical knowledge is the most important resource in the post-industrial era. With regard to the increasing importance of knowledge in a company, appropriate

10 Human Resource and Knowledge Management

221

mechanisms for maintaining and developing of this resource in companies must be successfully established (Blackler 1995).

10.6 Qualification Analysis and the Berufswertigkeit Concept (State-of-the-Art) In the logistics industry, the access for everyone—and in particular the career changers—for continuing education should be improved. Furthermore, the logistics industry requires specialists, meaning that the access to specialized personal skills also has to be simplified. One major aspect is the demographic change in Germany, which has to be counteracted with more flexible continuing education offers in which the practical on-the-job experiences of employees should be acknowledged. All these arguments call for a competence measurement concept, which is precise, practical and compatible in every economic sector and in particular the logistics sector. One measurement concept, which was developed since 2007, is the German “Berufswertigkeit” concept (Klumpp 2007; Klumpp et al. 2011). It fulfills the requirements of a general competence measurement instrument as it is connected to the concept of employability. The main idea of “Berufswertigkeit” is a concept of competence measurement of persons with different education degrees. The criteria for an effective competence measurement regarding demands of real-world companies and work processes are empirically evaluated and selected from business practice (Klumpp and Schaumann 2007). With these criteria, persons with different education backgrounds and degrees can be objectively compared while the results are output-oriented (no input and curriculum analysis and comparison but competence outputs of different qualification measures). It includes 36 qualification requirement criteria that represent the modern daily work which are used to individually measure (on a scale of 1-best to 5-worst) and calculates the aggregate “Berufswertigkeits index”. The qualification criteria are (Klumpp 2016): • • • • • • • • • • • • • •

Efficiency Independence and own initiative Flexibility and adaptability Work virtues Stress resistance Motivation and ability to lifelong learning and maintain own competence profile Coordinate work- and lifetimes Creativity Loyalty Risk-taking Charisma Ability to write and speak in German Knowledge of foreign language Ability to apply modern information- and communication technologies

222

• • • • • • • • • • • • • • • • • • • • • •

M. Klumpp et al.

Communication and rhetoric Assertiveness International and intercultural competence Customer focus Skills in mathematics and statistics Preparation of cost estimates and quotations Planning, implementation and documentation of orders and projects Negotiations capacity Analytical problem-oriented work Quality management (optimization of processes and products/service quality) Conceptual and strategic implementation of industry-specific knowledge Identification with the company Strategic orientation Understanding solutions of complex technical problems Basic knowledge of business administration Perception of functions of management and organization Conceptual analysis and work Planning and control of procurement and logistics processes Staff requirements and staff mission planning/staff development Team, staff and leadership Improving responsible care Legal knowledge

The Berufswertigkeits index (BWI) value is calculated by a summed and unweighted index of an individual personal evaluation of all the 36 qualification requirement criteria. The value range of the index begins by 0 and ends at 100.3 In this way, the output-oriented measuring concept “Berufswertigkeit” serves as a basic field-evaluation concept for the development of e.g. a European Qualification Framework for the logistics industry and integrates the required investigation of competences. A field survey with 1.068 persons from the German logistics industry to be evaluated by this concept was conducted in 2012. It was executed as a telephone survey in the German states of North Rhine-Westphalia and Hessen. Both states have a significant logistics industry cluster environment, i.e. around the inland port of Duisburg (North Rhine-Westphalia) and around the airport of Frankfurt (Hessen). In that survey, existing skills and competences of persons in the logistic industry are described. Additionally, traditional formal degrees in vocational and academic education can be classified according to evaluated practical competence levels. The following example (compare Fig. 10.7) outlines just a short example of the value and implications of a competence evaluation instrument such as the Berufswertigkeit Index. In this case, Berufswertigkeit Index results are depicted as a distribution function for different classes of achieved values in the BWI value. Alas, the first class on the left side of the graph indicates the share of people within the entire population 3 With “0” standing for an evaluation of all criteria with “poor”; and “100” representing an evaluation

of all criteria with “very good”.

10 Human Resource and Knowledge Management

223

Fig. 10.7 Example evaluation with the Berufswertigkeit concept (Klumpp et al. 2011)

who have achieved a (low) BWI value of under 45%. This is followed by the second class of people who have sustained a BWI between 45 and 50% and so on. The last group contains all persons of a specific group who score a (high) BWI of between 95 and 100%. If we now take a specific group as for instance a representative group of university graduates, some sort of “normal distribution” is usually obtained like the orange line in Fig. 10.7—as competence generally is quite “normally distributed” among a given group of people. This in itself is not so much of interest (besides the possible personal and management question that pertains a single person). However, of more interest and value is the comparison of different groups as outlined above: the group of graduates (orange) is compared to a representative group of senior managers (who usually hold a graduate degree, but are older and have additionally a long duration of practical management experience). So in-between the two groups it may be recognized that especially in the very high-level groups of BWI values, the senior management group has significantly larger shares in the high-level value group. As indicated by the arrow and the box in the figure, this indicates as difference or gap between the groups that obviously senior managers have gained competence during their career.4 This would indicate a quantifiable effect 4 Beware that this is just a “likelihood” match as in this case it was not a longitudinal study—therefore

not the same identical persons 20 or 30 years later—and could contain biases such as a timebias in academic qualification. So if, say, the senior managers today had their academic training about 25 years ago and this was significantly better than the academic education today, the effect seen in the figure may not be valid for an effect of informal learning. Nevertheless, it may just represent the difference in academic training quality between then and today. More work is needed to further establish such research results—but the cornerstone of outlining such competence evaluation concepts as the “Berufswertigkeit” concept is laid out.

224

M. Klumpp et al.

of informal learning; and this is state of the art research, as a quantifiable effect of such informal learning and competence effects has not yet been widely established. However, the effects on HR and knowledge management can be imagined: training measures will become measurable, so this is a topic for future HR research. The results and insights for the situation in Germany is readily transferable to other areas in high-level industry and service-oriented countries as competence requirements as well as education and training systems are highly uniform, at least in European and North-American countries, increasingly also in Asia.

10.7 Innovation in HR and Knowledge Management: Edugaming Earlier, the distinction between e-learning 1.0 and 2.0 have shown the evolution, which has taken place in this field recently. However strong that progress may be, it does not clear the approach from principal difficulties concerning implementation, or rather sustainable usage. This is due to issues inherent in (corporate) e-learning as it is understood here: being used as the vehicle for mandatory instructions (e.g. for new employees), motivation to take part in such measures is strongly external and seizes as soon as material or disciplinary incentives wane. Thus, measures which foster intrinsic user motivation are called for. Given the primary requirement of ongoing qualification in logistics to prepare employees effectively for the use of new technology and organization concepts, two initial propositions can be made: • Ongoing qualification needs to ensure efficient use of established workplace technology. Competitive advantages that result from specific technologies can only be exploited fully if qualification is made to measure—‘tools are only as good as the people who use them’. As much as technological innovations are celebrated, employee qualification (insofar as interaction takes place) has to match their pace. Every step ahead on technological grounds has to be met with one in user qualification. • Ongoing qualification provides potential for innovation. Only individuals and organizations appropriating the state of the art in relevant technology or media literacy, thus keeping close to the education innovation frontier, can be expected to generate novel ideas and concepts. Regarding logistics, the term ‘Pervasive Computing’ (computing happening anyplace, using any device, at any time, see e.g. Lucke and Rensing 2014) represents important current challenges, as it changes both methods of learning and teaching as well as individual learning behavior. In fact, the very definition of learning, its goals and assessment is challenged. With a view to the issue of skilled labor shortage, again, two perspectives are distinguished. First, there is the quantitative view, having as its focus the number of employees with a given set of skills. The second treats the issue as one of quality and is thus concerned with matters of competencies and ongoing qualification, thus with changes in employees’ set of skills. These views are linked to a dual solution concept,

10 Human Resource and Knowledge Management

225

which combines an expansive and an intensive approach: one may be concerned with, for instance, raising attractiveness of a given field of work (expansive), or with targeted efforts at ongoing worker qualification (intensive). With demographic change as a backdrop, the weight shifts continuously towards intensive approaches. Of these, one example is training on the job with educational games. We will provide a description of the concept of edugaming based upon a case from a recent research project. Recent efforts in this direction include gamification concepts for intralogistics, being tested in a lab-setting and with a focus on employee motivation and initial training-time reduction (GameLog, Munich Technical University, MIT Beer Game, Mortgage Service Game), the latter also being the main goal of the instructive software for picking, PickNick (Fraunhofer IML), or a plethora of e-learning concepts used for instance by arvato, DHL and Dachser, some of which having been developed by specialized firms such as the TÜV Rheinland Academy Group. Similar approaches are followed with the basic’ 10 Principles of Materials Handling’ in the warehousing and materials handling field. Research efforts within the scope of the project ‘MARTINA’ encompass the development of a smartphone-based edugaming-app as well as related efforts towards defining a topical map for ongoing qualification in logistics. This approach ensures that the resulting edugaming-app will be relevant and useful for blue- and whitecollar employees. Further benefits are transferability of game concepts to multiple upcoming qualification topics. While the research subject of topic identification is important in itself, we will only look at the issue of edugaming in logistics here. For the realization of the Edugaming concept a serious-game approach had been chosen by the developing team. The terms serious game, gamification, edutainment are often used interchangeably, while a clear definition is still debated (see e.g. Deterding et al. 2011). Nonetheless, concerning the intersection of these terms, one can state that they represent a connection of reality and game for conveying problem solving-skills and knowledge. The defining property of serious games is in a transfer of technology (development and design) from the arena of entertainment to that of education. Edutainment, then, in its original meaning, is a label for applications, which contain educational material embedded in game elements. However, the latter only function as rewards for the completion of lessons or study material. Earlier, we stressed the importance of intrinsic motivation, especially in cases where particular educational goals are pursued. Another related concept, widely stressed within gaming contexts, is immersion—which in its ideal state means an experience of one’s own physical presence in a virtual world. Intuitively, one might assume that both phenomena work hand-in-hand in an ideal educational game, with the educational content then being conveyed in passing. The suitability of this view is debated, especially with respect to the role of immersion (Hamari et al. 2016). Even if one’s ambitions in developing educational games are much more modest, these two concepts can be seen as guidelines. To develop a more comprehensive set of principles for educational games, some of the many theories of motivation provide a valid starting point. The following sketch of guiding principles for the design of educational games draws from the idea of intrinsic motivation with the aid of the widely known flow-concept (Csikszentmihalyi 1990), as well as from

226

M. Klumpp et al.

self-determination-theory (Ryan and Deci 2000). Flow describes a state of ‘optimal experience’, resulting from intensive participation in actions valued as enjoyable. It represents a state of supreme focus, affected neither by stray thoughts or worries, nor by consciousness of time—activities conducted solely for their own sake best account for this perception. For a gaming concept, a balancing of incremental challenges and skill-development appears central. The flow-concept, as described by Csikszentmihályi, assumes that a correct balancing of these influences ensures that experience remains in a corridor ‘above’ boredom, simple relaxation, and ‘below’ worry, anxiety, or arousal. For an educational game, this translates into requirements for e.g. difficulty, role of chance events, and all kinds of reward mechanisms. With regards to this concept, we list a few first attributes an ideal educational game should have: • • • •

Clear goals, to be achieved with reasonable effort Matching of difficulty of tasks and user abilities Causing an impression of being in control Requiring attention and focus to a degree that excludes simultaneous completion of other tasks • Direct, immediate feedback • Time perception is kept to the sidelines. The role of rewards and incentives (not necessarily external) for motivation, and especially for motivation to take up educational games, is crucial. While the importance of intrinsic motivation is beyond doubt, we may ask how the latter is achieved in an educational games context. In self-determination-theory, three needs have been stated as conditional to motivation (Nicholson 2015), all to be evaluated subjectively from an individual user’s perspective: • Competence, understood here as the impression of having gained knowledge or abilities to an extent that matters in applications; • Autonomy, which is congruence of behavior and identity as perceived by a user, thus closely related to freedom of choice; • Relatedness, a perception of involvement: this may be as part of a user community or multiplayer environment, whether its setup is competitive or not. Approaches at most comprehensive frameworks of game design elements can be found in Nicholson (2015) and Hunicke et al. (2004). There, the central role of narratives, aesthetics and sophisticated reward mechanisms is explained, as well as the ideal type of gamification, which would represent the creation of a space within which users would be able to establish and change rules and restrictions, as well as narratives, on their own. Our understanding of edugaming necessarily encompasses a subset of that framework, nevertheless putting emphasis on the aspects of • Intrinsic user motivation, even with the requirement of relating workplace content; • Exposition of users to narratives which are easily related to their work environment; • Nonlinear design and transferability of specialized content to all users.

10 Human Resource and Knowledge Management

227

Further, the entire application, once developed, is understood as a toolbox in the sense that its body, i.e. mechanics and aesthetics may be transferred to cover topics outside the field originally focused on (logistics). Of course, within a particular software development project, these (ideal) requirements form only part of the conditions, which encompass stakeholders, schedule, funding, specific research questions etc. Within the MARTINA project, the elements intrinsic motivation/immersion, exposition/narrative/nonlinear design, and transferability have been translated into project goals and design schemes. The educational game sets out from an individual perspective of an employee, having to accomplish tasks in cargos securing and customer care. In succeeding stages, tasks expand to different topics (for instance, dangerous goods) while growing user experience is mirrored in advances within a career in logistics. This progress is represented in a geographically growing logistics network the user manages. Further reward mechanisms are multi-dimensional, providing incentives for users to keep up a sustainable balance of cost, quality and time while playing games from the main storyline as well as optional ones. For the game design process, motivating to ‘pick up and play’ is a guiding thought, as is rewarding replay of optional stages with a finely graduated score in topics requiring diligence in practice (e.g. cargo securing). As is visible in the screenshots (Fig. 10.8), as a

Fig. 10.8 MARTINA app prototype, cargo-game (l.) and in-app progress (r.)

228

M. Klumpp et al.

general theme for graphical design, flat 2.0 has been chosen because its combination of reductionism and usability is well-suited to mobile devices, as the MARTINA App is primarily geared towards smartphones and tablets.

References Abdih Y, Joutz F (2005) Relating the knowledge production function to total factor productivity: an endogenous growth puzzle. IMF working Paper 05/74. IMF, New York Bell D (1973) The coming of post-industrial society. Basic, New York Berufswelt L (2017) Berufe in der Logistik [Professions in Logistics]. http://www.berufswelt-logi stik.de/?weiterbildung_logistik. Accessed 09 Feb 2017 Blackler F (1995) Knowledge and work and organizations: an overview and interpretation. Organ Stud 1021–1046 Blees I, Rittberger M (2009) Web 2.0 learning environment: concept, implementation, evaluation. eLearn Pap 3(15) Cervero RM (1988) Effective continuing education for professionals. Jossey-Bass, San Francisco Csikszentmihályi M (1990) Flow: the psychology of optimal experience. Harper & Row, New York Dabbagh N, Kitsantas A (2012) Personal learning environments, social media, and self-regulated learning: a natural formula for connecting formal and informal learning. Internet High Educ 15(1):3–8 Deterding S, Dixon D, Khaled R, NackeL (2011) From game design elements to Gamefulness: defining “Gamification”. In: Proceedings of the 15th international academic mindtrek conference: envisioning future media environments (MindTrek’11). ACM, New York, NY, USA, pp 9–15 Deutscher Bildungsrat B (ed) (1970) Empfehlungen der Bildungskommission, Strukturplan für das Bildungswesen. Eigenverlag, Bonn Eisenhardt KM, Santos F (2002) Knowledge-based view: a new theory of strategy? In: Pettigrew A, Thomas HWR. (eds) Handbook of strategy and management. Sage, London, pp 139–164 Esper T, Fugate B, Davis-Sramek B (2007) Logistics learning capability: sustaining the competitive advantage gained through logistics leverage. J Bus Logist 28(2):57–82 European Centre for the Development of Vocational Training (ed) (2011) The benefits of vocational education and training. Publications Office of the European Union, Luxemburg Hakkarainen K, Palonen T, Paavola S, Lehtinen E (2004) Communities of networked expertise: professional and educational perspectives. European Association for Research on Learning and Instruction, Amsterdam, Oxford Hamari J, Shernoff DJ, Rowe E, Coller B, Asbell-Clarke J, Edwars T (2016) Challenging games help students learn: an empirical study on engagement, flow and immersion in game-based learning. Comput Hum Behav 54:170–179 Hanft A, Knust M (2009) Continuing higher education and lifelong learning—an international comparative study on structures, organization and provisions, Oldenburg, München Hecker A (2012) Knowledge beyond the individual? Making sense of a notion of collective knowledge in organization theory. Organ Stud 33(3):423–445 Hofstein A, Rosenfeld S (2008) Bridging the gap between formal and informal science learning. Stud Sci Educ 28(1):87–112 Hunicke R, LeBlanc M, Zubek R (2004) MDA: a formal approach to game design and game research. In: proceedings of the challenges in games ai workshop, 19th national conference of artificial intelligence, pp 1–5 Jarvis P (1995) Adult and continuing education: theory and practice. Routledge, New York Klumpp M (2007) Begriff und Konzept Berufswertigkeit. FOM Working Papers No. 5 (07/2007), FOM, Essen

10 Human Resource and Knowledge Management

229

Klumpp M (2016) Logistics qualification: best-practice for a knowledge-intensive service industry. In: Zijm H, Klumpp M, Clausen U, ten Hompel M (eds) Logistics and supply chain innovation, bridging the gap between theory and practice. Springer, Heidelberg, Berlin, pp 391–411 Klumpp M, Schaumann U (2007) Requirements for leadership personnel and the concept Berufswertigkeit. Kölner Blätter für Wirtschaftspädagogik 12:3–50 Klumpp M, Clausen U, ten Hompel M (2013) Logistics research and the logistics world of 2050. In: Clausen U, ten Hompel M, Klumpp M (eds) Efficiency in logistics. lecture notes in logistics. Springer, Heidelberg, pp 1–6 Klumpp M, Kriebel K, Beschorner H, Buschfeld D, Dilger B, Diart M (2011) Berufswertigkeit konkret. Eul, Paderborn Lam A (2000) Tacit knowledge, organizational learning and societal institutions: an integrated framework. Organ Stud 2000:487–513 Lancioni R, Forman H, Smith MF (2001) Logistics and supply chain education: Roadblocks and challenges. Int J Phys Distrib Logist Manag 31(10):733–745 Liu X (2009) Linking competence to opportunities to learn: models of competence and data mining. Springer, New York Lucke U, Rensing C (2014) A survey on pervasive education. Pervasive Mob Comput, Spec Issue Pervasive Educ 14:3–16 Mangan J, Christopher M (2005) Management development and the supply chain manager of the future. Int J Logist Manag 16(2):178–191 Marsick VJ, Watkins KE (2001) Informal and incidental learning. New Dir Adult Contin Educ 89(1):25–34 Meyer B, Sugiyama K (2007) The concept of knowledge in KM: a dimensional model. J Knowl Manag 11(1):17–35 Mezirow J (1991) Trans Dimens of Adult Learn. Jossey-Bass, San Francisco Michaels E, Handfield-Jones H, Axelrod B (2001) The war for talent. Harvard Business School Publishing, Boston Nicholson S (2015) A RECIPE for meaningful gamification. In: Reiners T, Wood LC (eds) Gamification in education and business. Springer, Cham, pp 1–20 Nonaka I (1994) A dynamic theory of organizational knowledge creation. Organ Sci 5(1):14–37 Nonaka I, Takeuchi H (1995) The knowledge creation company. How Japanese companies create the dynamics of innovation, Oxford Probst G, Raub S, Romhardt K (2006) Wissen managen—Wie Unternehmen ihre wertvollste Ressource optimal nutzen. Gabler, Wiesbaden Polanyi M (1985) Implizites Wissen [The tacit dimension, 1966]. Suhrkamp, Frankfurt a.M Rao K, Stenger AJ, Wu H-J (1998) Integrating the use of computers in logistics education. Int J Phys Distrib Logist Manag 28(4):302–319 Rogers H, Braziotis C (2016) Current issues in teaching logistics management. In: Kotzab H, Pannek J, Thoben KD (eds) Dynamics in logistics. Springer International, Cham, pp 553–563 Ryan RM, Deci EL (2000) Intrinsic and extrinsic motivations: Classic definitions and new directions. Contemp Educ Psychol 25:54–67 Worldbank (2017) Logistics performance index global ranking. http://lpi.worldbank.org. Accessed 09 Feb 2017 Wu YC-J (2007) Contemporary logistics education: an international perspective. Int J Phys Distrib Logist Manag 37(7):504–528 Zijm H, Klumpp M (2016) Logistics and supply chain management: developments and trends. In: Zijm H, Klumpp M, Clausen U, ten Hompel M (eds) Logistics and supply chain innovation, bridging the gap between theory and practice. Springer, Heidelberg, Berlin, pp 1–19

Part IV

Functions in Production and Logistics

Chapter 11

Inbound Logistics Stefan Minner

Abstract In this chapter, we discuss selected concepts and decision support models for inbound logistics. Inbound logistics comprises all activities that secure supply for manufacturing, assembly, and retail operations. The associated information and materials flow involve different strategic and operational decisions that will influence transportation, handling and inventory costs. They depend on various parameters, such as the variety and volumes of material requirements, the supplier base and respective locations. Section 11.1 introduces the basic concepts, key performance indicators, and problem parameters. Section 11.2 presents advanced decision support models for the design, implementation and operation of the concepts. Section 11.3 includes a case study. Selected state-of-the-art research and recent advances are discussed in Sect. 11.4.

11.1 Concepts of Inbound Logistics (Basic) Logistics is integral for operating global supply chains. Therefore, the design and operations of supply chains as part of an overall operations strategy needs to consider the planning of locations and capacity at the strategic level and transportation and inventory management at the operational level. Transportation logistics connects the different levels of a supply and distribution network. A typical transportation process that connects two locations in the network consists of the steps of loading, pre-carriage, main haul, on-carriage, and unloading. In this chapter, we will focus on inbound logistics, i.e. on the design and operations of incoming material flows and the associated information (and financial) flows. A typical automotive final assembly plant such as the one at BMW in Leipzig, Germany, produces about 800 cars each day and receives some 500 truck deliveries from its suppliers per day. Because of these large volumes, the optimization of the inbound transportation with high volume and high product variety is mandatory if we wish to minimize cost and achieve logistics efficiency. Strategically, the delivery conS. Minner (B) School of Management, Technical University of Munich, Munich, Germany e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_11

233

234

S. Minner

cept, including supplier and parts segmentation, needs to be decided. Operationally, scheduling and sequencing to avoid queues and waiting times for the carriers and a smooth capacity utilization represent challenging planning tasks. For the design and operation of inbound logistics processes, we strive for achieving efficiency, but at the same time want to allow for flexibility and reactiveness. Securing a steady and reliable supply is core to manufacturing and sales processes, in particular when low inventories are kept to buffer against variability and volatility. When environmental conditions change either gradually or as a result of disruptions, efficiently organized manufacturing and assembly processes, such as those that we find in the automotive industry, might face significant challenges. Just-in time (JIT), just-in-sequence (JIS) supply systems for materials are very common in the automotive industry. Suppliers locate at a certain distance from the OEM plants to be able to secure a continuous and stable delivery of materials. The performance of inbound logistics can easily be at risk when required environmental conditions change significantly due to disruptions. During the refugee crisis in Europe in 2016, the vulnerability of these JIT and JIS systems became obvious. Border controls were reintroduced by Austria in September. This resulted in longer waiting times at the border to Germany. If such controls were in place permanently, OEMs and suppliers would need to consider reorganization of the JIT delivery processes between Eastern European supply plants and OEM plants in Germany. Prognos expected significant losses for European economies as a consequence of changes to the Schengen Agreement, which allows for free border crossing between the countries that signed the agreement (Prognos 2016).

11.1.1 Definition and Performance Criteria Inbound logistics comprises all activities that secure the supply for manufacturing and assembly or sales. These activities range from order placement and order allocation between suppliers to a chosen delivery and transportation concept for the receipt and storage or immediate use of the materials. When multiple plants or warehouses of a company and multiple suppliers are involved, we consider a many-to-many logistics system; when multiple suppliers deliver to a single warehouse, it is a manyto-one system (see Daganzo 2005). The large amount and increasing variety of goods received by plants and warehouses led to different delivery and transportation concepts that will be introduced in the following. The evaluation of and choice between concepts is typically based on the following key performance indicators: • Transportation costs • Handling costs

11 Inbound Logistics

235

• Inventory costs • Service level agreements For transportation, we distinguish between full truckload (FTL) and less-than truckload (LTL) deliveries. One trade-off to be solved when deciding between these two options is the involved transportation frequency, i.e. more frequent deliveries require less inventory but might not lead to full truckloads, whereas less frequent deliveries result in inventories both at origins and destinations, but they use transportation capacity efficiently. In the first case, incoming trucks need to be scheduled and sequenced. In the second case, delivery concepts aim at overcoming the problem by consolidating LTL deliveries from multiple suppliers. For inventory replenishment, synchronization of demand and supply is typically found in just-in-time systems with frequent transportation and low inventory, whereas batch ordering and replenishment to stock is another option with larger inventories but lower transportation quantities. Incoterms (International commercial terms) are standardized terms between the supplier and the buyer that specify the responsibility to organize the transportation, with delivery ex-works, EXW, or delivery and duty paid, DDP, as the two extremes where full responsibility is with the buyer in the former case and with the supplier in the latter case. Besides transportation, these standardized terms further specify at which point of the transportation chain risks transfer from the supplier to the buyer and who is responsible for paying taxes, insurance, and duties.

11.1.2 Delivery and Transportation Concepts In the following, we introduce the main delivery and transportation concepts, along with their advantages and disadvantages. The delivery concept distinguishes between direct delivery from a supplier to a consumption point and the delivery to a warehouse for storage, regrouping and onward transportation. Just-in-time (JIT) and justin-sequence (JIS) systems are widely used direct delivery concepts. For example, the BMW plant in Leipzig receives 30% of its supplies by direct delivery for various modules to 36 different delivery points within the plant (Klug 2010). This is a typical example of high volume, high frequency JIT/JIS. It avoids high inventories in environments with a large product variety by superior organization concepts, but requires significant organization, coordination, and reliability of inbound logistics processes. Figure 11.1 schematically illustrates a direct delivery structure with four supply factory origins and one destination factory. Where necessary, warehouses indicate intermediate handling nodes. Just-in-time implies that materials are delivered very close to the time of consumption, thereby avoiding inventories and reducing lead times. Just-in-sequence is an organizational concept that also delivers a variety of materials in the sequence required for further processing or assembly. However, JIT/JIS logistics requires stable volumes and sequences and is therefore typically limited to supply from geo-

236

S. Minner

Fig. 11.1 Direct delivery

Fig. 11.2 Consolidation hub

graphically close supplier locations. The main advantage of direct delivery is its simplicity in coordination and the reduction of inventories for the buyer. A disadvantage may be that no full truckloads can be used when less-than-truckload volumes are required within an allowed time frame, together with the vulnerability to disruptions. In Sect. 11.2, we present a simple decision model for finding the transportation frequency for direct delivery that resolves the trade-off between inventories at origin and destination on the one hand and capacity utilization, i.e. full truckloads, on the other hand. A major disadvantage of direct deliveries is a potentially large number of transportation links between many suppliers and many destinations, which results in a low utilization of transportation resources when volumes do not allow for full truckloads. Therefore, another concept for inbound logistics with the aim of coordinating freight consolidation has been introduced. It utilizes consolidation hubs as shown in Fig. 11.2. Here, suppliers do not have sufficient volume or delivery frequencies are too low for direct deliveries with full truckloads. To avoid a number of poorly utilized shipments and to achieve further economies of scale through the use of higher capacity transportation resources, smaller shipments can be collected at consolidation hubs and then jointly shipped to the destination. The main advantage of this approach is a higher utilization of transportation capacities; however, the concept causes additional coordination and handling costs for carrying out the consolidation. In a milk-run system, less-than truckload deliveries from multiple suppliers are collected by a truck to achieve a higher fleet utilization and continuous supply. As opposed to the consolidation system, a truck consecutively visits several supplier locations to pick up goods, whereas the consolidation system still involves multiple (smaller) trucks for shipments between the suppliers and the consolidation point and larger capacities between the consolidation center and the destination. As a

11 Inbound Logistics

237

Fig. 11.3 Milk-run system

Fig. 11.4 Cross docking system

disadvantage, more loading operations and coordination of routing are necessary. A milk-run system results in a vehicle routing problem for the pickups or, if multiple suppliers and destinations are involved, in a combined pickup and delivery problem (see Toth and Vigo 2014). If combined with the pickup frequency at suppliers, we face a special form of an inventory routing problem (see Coelho et al. 2014). Besides the plant-milk-run as shown in Fig. 11.3, another option is to do milk-runs starting and ending at consolidation points, by potentially using smaller vehicles, then consolidate the collected volumes and to send full truckloads using transportation means with even larger capacities to the common destination. When multiple suppliers service multiple plants, an important logistics concept in many-to-many distribution is cross-docking, which is also widely used in retail transportation logistics. Multiple suppliers deliver pre-packed goods to a cross-docking point where the packs are routed to the outbound deliveries for multiple destinations. A coordinated arrival from the suppliers and departures to the destinations reduces inventories and at the same time enables full truckloads, both at the inbound and outbound stages. While each supplier only ships less-than-truckload volumes to each customer plant or warehouse, aggregating the shipments for multiple destinations allows full truckloads to the cross dock and, after sorting and re-grouping, again full truckloads from the cross-dock to the multiple destinations. This inbound concept, however, requires more coordination efforts and handling processes (Fig. 11.4).

238

S. Minner

Fig. 11.5 Hybrid system with direct and consolidated shipments

Remaining smaller/smallest deliveries are typically handled through carriers using groupage concepts, which usually need to allow for larger lead times so that these providers can bundle shipments along with other business. Due to the size of supply networks and the involved product variety in practice, a pure concept as illustrated above is seldom found. Instead, hybrid systems as shown in Fig. 11.5 combine the advantages and avoid the disadvantages by segmenting suppliers and products. For example, suppliers with large shipment volumes can easily fill a single container or several containers every week (without loss of generality assuming this is the intended shipping frequency) and directly supply the customers’ manufacturing plants. Suppliers with a smaller weekly volume will deliver to a consolidation point where full containers or trucks are generated and then forwarded to the customer. The segmentation of suppliers and/or products shipped directly or via a consolidation hub represents a non-trivial optimization problem that first needs to allocate suppliers or individual products of a supplier to a certain shipment mode and secondly must build shipment units to satisfy loading constraints. One instrument to support such segmentation is ABC analysis. In a typical ABC classification, the A-category includes suppliers with high volumes that can be organized with direct deliveries at high frequencies, i.e. multiple trucks every day. In category B, less-than truckload shipment volumes require some consolidation either via consolidation hubs or through being organized as milk-runs, whereas category C includes small shipments operated through third-party carriers in groupage.

11.2 Planning and Decision Support for Inbound Logistics (Advanced) In this section, we present decision support for strategic, tactical, and operational problems in inbound logistics.

11 Inbound Logistics

239

11.2.1 Network Design The network design problem for the concepts introduced in the previous section includes several choices. The number of stages, the number of facilities at each stage and their location, as well as capacities and the adjustment of facilities and capacities over time need to be determined. Simultaneously, the delivery and transportation concepts are defined, ideally even anticipating the operational decisions and cost impact (see Schneeweiß 2003 for a hierarchical and distributed decision making framework). Melo et al. (2009) give an overview on supply chain design models. With regard to the above delivery and transportation concepts, this means the decision about the number and location of consolidation centers and cross docks and the optimization of FTL and LTL transportation and routing operations.

11.2.2 Inventory and Freight: Frequency Optimization Next, we introduce a simple transportation frequency optimization problem for inbound logistics that builds on the assumptions of the Economic Order Quantity (EOQ) model. It illustrates the trade-offs between inventories and transportation costs and also investigates full truck loads versus economic load factors (see Burns et al. 1985). Assume a single link supply chain with a supplier, a buyer and direct delivery. Over an infinite planning horizon in continuous time, N products are produced at the supplier at rate pi and consumed at the buyer at rate di (i = 1, . . . , N ). Backorders are not permitted. For long-term stability, we require di = pi . Goods stored at the supplier and the buyer are subject to inventory holding costs per unit and unit of time of h is and h id . Every unit of product i requires ai units of transportation capacity. The single available truck has a capacity of W units and every shipment between the supplier and the buyer causes fixed costs of c. For simplicity of exposition, we assume negligible transportation times and empty detours free of charge. The decision to be taken is the optimal time between two consecutive shipments T that minimizes the average cost per unit of time. To satisfy the demand between consecutive shipments, the shipment quantities qi for each product have to equal demand during that period of time, i.e. qi = di T . Figure 11.6 illustrates the inventory development for a single product at the origin (supplier) and the destination (buyer), with both of them following the well-known saw-tooth pattern of economic lot-sizing models. At the source, inventory is zero after truck departure and increases with the production rate pi until the next departure. At the destination, inventories of all products equal qi = di T after truck arrival and deplete at rate di until the inventory level reaches zero and the next truck will arrive. Under these assumptions, the cost function is given as follows. There are fixed costs per shipment of c, i.e. transportation costs per unit of time are Tc . The average inventory holding cost of every product, including origin and destination, (h s +h d )d T is i 2i i . Note that this holding cost assignment uses a centralized perspective,

240 Fig. 11.6 Inventory development at origin and destination

S. Minner Inventory

Source

T

2T

Inventory

Time

DesƟnaƟon

T

2T

Time

i.e., includes both the impact on the supplier’s and the buyer’s cost. When taking a decentralized perspective for a single entity, one can set one of the two holding cost parameters to zero. The average cost per unit of time as a function of the time between two shipments then becomes N 1 s c C(T ) = + (h + h id )di T. (11.1) T 2 i=1 i Figure 11.7 visualizes the (total) average cost per unit of time as a function of the time T between two shipments and its two components, inventory holding costs and transportation costs per unit of time. When we increase the time between shipments, i.e. we decrease the transportation frequency, holding costs increase linearly due to the storage of more units at the origin and destination while transportation costs decrease. For the numerical values, we assume the following example data. The cost per truck going from origin to destination with capacity W = 66 is c = 500. We consider i = 1, 2 products with demand and supply rates d1 = 10 and d2 = 15 and define holding costs h i = h is + h id using values h 1 = 5 and h 2 = 3, and transportation capacity consumptions of unit size. The transportation capacity constraint is N 

ai di T ≤ W,

(11.2)

i=1

i.e., the minimum of either the time needed to fill the truck or the economic optimum. The above optimization problem minimizes a convex objective function under a linear constraint. Using a Lagrange multiplier approach, the optimal time between two shipments is given by

11 Inbound Logistics

241

1000

Total cost TransportaƟon cost

800

Cost

Holding cost

600 400 200 0

0

1

2

4

3

5

6

7

8

9

10

Time T Fig. 11.7 Cost trade-offs and transportation frequency

 ∗



W

T = min  N

i=1

ai di

,

N

2c

i=1

h i di

 .

(11.3)

For the introduced numerical example, the cost-optimal delivery time T ∗ is 3.24, which, however, is not feasible as a truck is already filled after 2.64 periods. Figure 11.7 further illustrates the well-known property that the cost-optimal solution of such a kind of lot-sizing model is rather robust, i.e. even significant errors in parameters that lead to larger deviations of the chosen transportation frequency from the optimal one only result in moderate cost increases as the total cost curve is very flat around the optimum. This basic model can be adapted to more realistic assumptions (see Burns et al. 1985 and Blumenfeld et al. 1985). Here, it serves to illustrate the basic trade-off between more frequent deliveries resulting in less inventories but higher transportation costs and the opposite scenario. The previous models assume deterministic demands. To cope with supply uncertainty, various strategies are available, the most popular being safety time and safety stock. A widely used (text-book) formula for setting safety stocks (SST) in an environment under stochastic demands and lead times, simultaneously assuming independent normally distributed random variables and a non-stockout probability constraint is given by:  (11.4) SST = k Lσ D2 + μ2 σ L2 −1 where k = F0,1 (α) is the required safety factor for achieving a non-stockout prob−1 ability α, F0,1 (x) denotes the inverse of the standard normal distribution, μ is the mean demand per period, σ D2 the variance of periodic demand, L the mean lead time, and σ L2 the variance of the lead time (see e.g. Silver et al. 2017).

242

S. Minner

11.2.3 The Joint Replenishment Problem Another type of decision support model that allows for more realistic assumptions in inbound coordination is the joint replenishment problem. The model assumes discrete time periods t = 1, . . . , T within a finite planning horizon of length T . Multiple products k = 1, . . . , K with dynamic demands dkt are considered. Backorders are not permitted, i.e. inventory levels at the end of the period have to be non-negative. As before, inventories are subject to holding costs h k for product k per unit per unit of time. Note that, in the following formulation we only take the perspective of the buyer, i.e., holding costs only include the buyer’s inventory. The main difference between joint replenishment problems and simple single-product lot-sizing models is the fixed cost structure, which for the joint replenishment problem includes a major setup cost A independent of the number of products included in the replenishment and minor setup costs Ak for each product replenished in a period but independent of the order quantity. The major setup cost therefore addresses the replenishment, e.g. truck delivery, whereas the minor setup costs address the handling and processes per product. The solution of the model coordinates the inbound logistics across products, i.e. which products to replenish together and at what frequency. The following mixedinteger linear program supports such joint and coordinated replenishments. Decision variables • • • •

qkt order quantity of product k in period t γt binary indicator if there is any (major setup) order in period t u kt binary indicator if there is an order (minor setup) for product k in period t ykt inventory level of product k at the end of period t, initial inventories yk0 are given

Optimization model min

T  t=1

(Aγt +

K 

(h k ykt + Ak u kt ))

(11.5)

k=1

s.t. ykt = yk,t−1 + qkt − dkt , t = 1, . . . , T, k = 1, . . . , K qkt ≤ u kt M, t = 1, . . . , T, k = 1, . . . , K u kt qkt

(11.6) (11.7) ≤ γt , t = 1, . . . , T, k = 1, . . . , K (11.8) ≥ 0, ykt ≥ 0, u kt ∈ {0, 1}, γt ∈ {0, 1}, t = 1, . . . , T, k = 1, . . . , K (11.9)

The objective function (11.5) minimizes the sum of major and minor transaction costs, as well as inventory holding costs for all products k and periods t. Constraint (11.6) is the inventory balance equation, which enforces that the final inventory at the end of a period is equal to the initial inventory plus delivered units minus demanded units. Constraints (11.7) and (11.8) represent logical constraints that ensure that an

11 Inbound Logistics

243

order quantity of a product can only be positive if the corresponding indicator is equal to one and the product-specific indicator itself can only be one if the major indicator is one. M defines a sufficiently large number. Non-negativity and binary variable constraints (11.9) complete the model. This mixed-integer-linear programming formulation can be solved by respective solvers. To do so, it might be advantageous to use a different model formulation, see Narayanan and Robinson (2006). While this basic formulation assumes dynamic but deterministic and therefore known demand, this assumption might not be realistic and requires extension, in particular for retail inbound logistics. For extensions, we refer to Minner and Silver (2005) and Minner (2009).

11.2.4 Inventory Routing Problems Inventory routing combines the two fundamental problems in logistics, inventory management and transportation. The basic dynamic multi-period single-product lotsizing model is combined with the vehicle routing problem. For a literature review, see Coelho et al. (2014). The following model formulation combines the two traditional mixed-integer linear programming models for lot-sizing and vehicle routing. All deliveries to customers i = 1, . . . , n originate from a single central depot i = 0. Customer demands dit for periods t = 1, . . . , T need to be satisfied, i.e. backorders are not permitted. Transportation between nodes i and j causes distance dependent transportation costs ci j and trucks of a homogenous fleet have a limited capacity of W . Inventories at customer i at the end of a period are subject to holding costs h i per unit and unit of time. In every period t, decisions are taken about whether to supply customer i (if γit = 1) or not (γit = 0), about supply quantities qit , non-negative inventory levels at the end of period yit , and vehicle routing variables xi jt that decide whether a truck goes from node i to j in period t. Decision variables • • • • •

γit delivery to customer i in period t qit delivery quantity to customer i in period t yit inventory level of customer i at the end of period t xi jt routing variable if truck goes from customer i to j in period t u it remaining capacity of a truck after supplying customer i in period t

The optimization model is

244

S. Minner

min

⎛ ⎞ T n  n n    ⎝ ci j xi jt + h i yit ⎠ t=1

i=0 j=0

s.t. yit = yi,t−1 + qit − dit , i = 1, . . . , n; t = 1, . . . , T n  xi jt = γ jt , j = 1, . . . , n; t = 1, . . . , T i=0 n 

(11.10)

i=1

xi jt = γit , i = 1, . . . , n; t = 1, . . . , T

(11.11) (11.12) (11.13)

j=0

qit ≤ Mγit , i = 1, . . . , n; t = 1, . . . , T

(11.14) u 0t = W, t = 1, . . . , T (11.15) u jt ≤ u it − qit + (1 − xi jt )M, t = 1, . . . , T, i, j = 0, . . . , n; i  = j (11.16) xi jt ∈ {0, 1}, i, j = 0, . . . , n, i  = j, t = 1, . . . , T (11.17) qit ≥ 0, u it ≥ 0, yit ≥ 0, γit ∈ {0, 1}, i = 1, . . . , n; t = 1, . . . , T (11.18)

The objective function (11.10) minimizes the sum of transportation and inventory holding costs. The constraints for every period t represent inventory balances (11.11) with yi0 denoting the initial inventory, truck arrival and departure at locations that require delivery during that same period (11.12) and (11.13), and logical constraints that limit supply quantities to those days that are scheduled for delivery (11.14). Loading capacity constraints and the avoidance of sub-tours are achieved through (11.15) and (11.16). As for the vehicle routing problem, several extensions (see e.g. Toth and Vigo 2014) are possible to this model, i.e. time windows and forbidden days, the combination of pickup and delivery when multiple suppliers deliver to multiple plants, etc. Turan et al. (2017) present a variable neighborhood search (VNS)-approach for a perishable (newsvendor-type) product with an option for resupplying stock once during the sales day. The inbound coordination problem is the combined routing, delivery timing and resupply quantity allocation problem.

11.3 Case Study An automotive supplier is concerned about its transportation cost spent for inbound logistics. A first analysis revealed that the current situation is far from best in class. The management therefore discusses the introduction of a state-of-the-art Transportation Management System (TMS) that makes use of the latest technologies in measuring, analyzing, visualizing, and optimizing world-wide transportation flows. In a first study, the following problem areas for improvement were detected: • Insufficient load factors of trucks supplying material to the plants

11 Inbound Logistics

245

• Unclear rules for deciding about direct shipments from large suppliers to plants • Significant waiting times of trucks when unloading at the plant warehouses • No company-wide guidelines for tendering transportation services • No clear rules for consolidation of shipments and deciding about transportation lot-sizes • A mix of multiple contracts with logistics service providers to pickup supplies even within a single region • No clear segmentation of suppliers and part numbers with regard to volume and frequency of pickup and supply When discussing available tools for the digitalization of inbound logistics, the responsible logistics managers criticize that the technology for monitoring and administering inbound transportation is only part of the problem. The effort necessary to parameterize such software systems in such a way that they can create and maintain a reliable and up-to-date data base is often underestimated and might offset the benefits achieved by increasing variability. There is also a lack of human resources that are able to design and execute data analysis capabilities and turn these into optimization benefits. The amount of available data would need more automation and careful analysis to find patterns for designing future tenders and to build regions for assignment. The question is further about the frequency at which to revise the inbound logistics network due to changing product generations, a different supply base, changing volumes, and increasing volatility.

11.4 Research for Inbound Logistics (State-of-the-Art) Increasing computing capabilities, data availability and advanced optimization methods allow for extended decision support beyond the simple models presented in the previous section. A recent trend in transportation optimization is the incorporation of uncertainty, i.e. to include fluctuations in volumes and/or transportation times. Initial system configurations, such as the locations of consolidation points and cross docks, need to be selected in a robust way, which will operationally allow for some flexibility when it comes to adjusting to changing environments. In retail, with its expanding multi-channel strategies, the load and capacity utilization changes considerably over time, exhibiting large peaks towards end of the year. This requires more flexible and adjustable logistics systems. Another problem at the operations-finance interface are multiple options to finance inventory. If inbound logistics is outsourced to logistics service providers, so can be the replenishment and financing involving a bank. When outsourcing inbound logistics, an important design problem is the duration of the contract and the organization of the tendering process. This, of course, depends on the selected inbound logistics strategy and the resulting transportation and warehousing service required.

246

S. Minner

In response to many supply chain disruptions recently observed, many proactive and reactive risk management strategies have been proposed, in particular strategies that use multiple and backup suppliers (Minner 2003; Yao and Minner 2017). While inventory and availability criteria have been considered in respective research, the impact on transportation and transportation consolidation still requires additional investigation, in particular when we consider multi-stage supply chains. Following the idea of data-driven optimization, i.e. using past demand data for optimization, rather than decoupling the planning problem into a forecasting and replenishment optimization problem, Taube and Minner (2017a) determine replenishment patterns for retail operations, i.e. they determine what products should be replenished on what days in order to guarantee availability. At the same time they smooth handling capacities at stores and a central warehouse.

11.4.1 The Loading Dock Waiting Time Problem One of the challenges in inbound logistics is the problem of coordinating truck arrivals and unloading to avoid waiting times at loading docks. The problem is caused by the uncoordinated arrival of trucks delivering material and the non-synchronized warehouse capacity at warehouse gates, for example because many carriers prefer deliveries in the early morning. This, according to the Bundesamt für Güterverkehr (2011), a federal agency for monitoring freight traffic in Germany, leads to considerable waiting times of 1–2 h per truck and warehouse. One solution that has been suggested for this problem are time slot management systems where carriers have to book certain delivery time windows at a certain fee (see Elbert et al. 2016). Providers of such solutions are for example Mercareon (www.mercareon.com) and Transporeon (www.transporeon.com). However, as this system primarily works as a first-come-first serve booking platform, the planning flexibility of carriers is limited by the slots that have to be booked ahead of time and might not be a good fit with their preferred truck routing. To improve the situation, various supply chain coordination mechanisms are available that need to be tailored to the problem at hand. A first approach is to share information with carriers about expected waiting times at each hour of the day, thereby giving them a better basis for their delivery planning. Research-wise, this information reduces the uncertainty about the stochastic service times for waiting and unloading in a stochastic vehicle routing problem. Lemke et al. (2014) show that sharing information can be an effective means by which to improve waiting times if enough, but not all, carriers pick up this information. A disadvantage if too many carriers use the information in the same way is that they all will adjust their plans towards other (the same) time windows and thereby not avoid but only shift the problem. A solution recently proposed in Karänke et al. (2015) is the use of auction mechanisms for allocating available warehouse unloading time slots to carriers following their bids for certain routes. Carriers can submit routes and bids to a clearance platform that then selects and awards proposed routes to the carriers, at the same time balancing the utilization at the warehouse gates. Different auction mechanisms for truthful bidding

11 Inbound Logistics

247

are presented and compared in a numerical experiment to show their effectiveness in achieving coordination between the carriers. Another way to solve the problem is the use multi-agent systems, similar to, e.g., the barge-terminal visit problems in the Port of Rotterdam (see the case in Chap. 27).

11.4.2 Sequencing and Resequencing In order to organize inbound logistics, trucks delivering material, as well as items being delivered for operations, require sequencing to achieve operational efficiency and to satisfy JIS constraints. On the one side, incoming trucks need to be sequenced in order to minimize waiting times, on the other hand, the utilization is increased at the loading docks. In a stochastic version of this problem where arrival and processing times are random, this system can be seen as and analyzed by queuing systems with a single or with multiple servers. In a deterministic version, it can be modelled as a scheduling problem for minimizing the cycle time to process all incoming trucks in minimum time (see Boysen et al. 2010). In just-in-time, just-in-sequence systems, the buyer orders a certain number of products for a certain time frame or with the next delivery truck. For high product variety environments, the components need to be provided in a sequence, which might not necessarily be the optimal one for the supplier to produce. In such cases, resequencing operations can be beneficial and there are various organizational possibilities for resequencing, storage and sorting to rebuild the original sequence (see Boysen et al. 2012). Taube and Minner (2017b) present an approach for effectively organizing the restoration of an original OEM sequence of parts that is produced in another sequence to achieve different efficiency goals at the supplier. The resequencing strategy is shown in Fig. 11.8, where an original sequence needs to be delivered just-in-sequence to the OEM. The different products can be resequenced, which essentially represents a sequencing problem using a travelling salesman formulation, with the additional constraint that, after production, the products can be put into different storage lanes and pulled from there to restore the required sequence without interim buffering. The assignment of products to lanes is essentially a vehicle routing problem, where the lanes represent the vehicles with loading constraints. Combining the two subproblems, the production sequence (TSP) and lane storage (VRP) can be formulated as a straightforward mixed-integer-linear program. For larger problems, Taube and Minner (2017b) suggest and evaluate a simple lookahead method that performs well with short computation times.

11.5 Further Reading Gudehus and Kotzab (2012) provide an exhaustive coverage of logistics. A comprehensive collection of material on inbound logistics can be found at the website www.inboundlogistics.com. For aspects of modeling and solving vehicle rout-

248

S. Minner

Fig. 11.8 Resequencing in JIS assembly (Taube and Minner (2017b))

ing problems, the interested reader is referred to the collection by Toth and Vigo (2014), while a broad coverage of inventory management aspects and references is provided in Silver et al. (2017). For more details on transportation markets and transportation data analysis, please see Ben-Akiva et al. (2013) and Washington et al. (2011). Sinha and Labi (2007) is a textbook reference for details on transportation performance evaluation, whereas Chandra and Grabis (2007) introduce concepts, solutions and applications for supply chain configuration in general. As for many other sectors, logistics is currently undergoing (disruptive) changes caused by the fourth industrial revolution (Industry 4.0). Developments and research requirements are summarized by Delfmann et al. (2017).

References Ben-Akiva M, Meersman H, Van de Voorde E (2013) Freight transport modelling. Bingley, Emerald Blumenfeld DE, Burns LD, Diltz JD, Daganzo CF (1985) Analyzing trade-offs between transportation, inventory and production costs on freight networks. Transp Res Part B: Methodol 19(5):361–380 Boysen N, Fliedner M, Scholl A (2010) Scheduling inbound and outbound trucks at cross docking terminals. OR Spectr 32(1):135–161

11 Inbound Logistics

249

Boysen N, Scholl A, Wopperer N (2012) Resequencing of mixed-model assembly lines: survey and research agenda. Eur J Oper Res 216:594–604 Bundesamt für G (2011) Sonderbericht zur Situation an der Laderampe. www.bag.bund.de/ SharedDocs/Downloads/DE/Marktbeobachtung/Sonderberichte/SB_Rampe.pdf Burns LD, Hall RW, Blumenfeld DE, Daganzo CF (1985) Distribution strategies that minimize transportation and inventory costs. Oper Res 33(3):469–490 Chandra C, Grabis J (2007) Supply chain configuration. Springer, Berlin Heidelberg Coelho LC, Cordeau JF, Laporte G (2014) Thirty years of inventory-routing. Transp Sci 48(1):1–19 Daganzo CF (2005) Logistics systems analysis, 4th edn. Springer, Berlin Heidelberg Delfmann W, ten Hompel M, Kersten W, Schmidt T, Stölzle W (2017) Logistics as a science central research questions in the Era of the fourth industrial revolution. BVL International, Bremen Elbert R, Thiel D, Reinhardt D (2016) Delivery time windows for road freight carriers and forwarders—influence of delivery time windows on the costs of road transport services. In: Commercial transport—proceedings of the 2nd interdisciplinary conference on production, logistics and traffic 2015. Springer, pp 255–274 Gudehus T, Kotzab H (2012) Comprehensive logistics, 2nd edn. Springer, Berlin Heidelberg Karänke P, Bichler M, Minner S (2015) Retail warehouse loading dock coordination by coreselecting package auctions. In: ECIS 2015 completed research papers. Paper 91 Klug F (2010) Logistikmanagement in der Automobilindustrie. Springer, Berlin Heidelberg Lemke J, Bichler M, Minner S (2014) Fighting waiting times at the retail warehouse: how information aids coordination. In: Proceedings of the 7th international scientific symposium on logistics, German logistics association (BVL) Melo MT, Nickel S, Saldanha-da-Gama F (2009) Facility location and supply chain management a review. Eur J Oper Res 196:401–412 Minner S (2003) Multiple supplier inventory models in supply chain management: a review. Int J Prod Econ 81–82:265–279 Minner S (2009) A comparison of simple heuristics for multi-product dynamic demand lot-sizing with limited warehouse capacity. Int J Prod Econ 118(1):305–310 Minner S, Silver EA (2005) Multi-product batch replenishment strategies under stochastic demand and a joint capacity constraint. IIE Trans Sched Logist 37(5):469–479 Narayanan A, Robinson EP (2006) More on “models and algorithms for the dynamic-demand joint replenishment problem”. Int J Prod Res 44(2):383–397 Prognos (2016) Abkehr vom Schengen Abkommen—Gesamtwirtschaftliche Wirkungen auf Deutschland und die Länder der Europäischen Union. Technical Report, Munich Schneeweiß C (2003) Distributed decision making, 2nd edn. Springer, Berlin Heidelberg Silver EA, Pyke DF, Thomas DJ (2017) Inventory and production management in supply chains, 4th edn. CRC Press, Boca Raton Sinha KC, Labi S (2007) Transportation decision making. Wiley, Hoboken Taube F, Minner S (2017a) Data-driven assignment of delivery patterns with handling effort considerations in retail. Working paper Taube F, Minner S (2017b) Resequencing mixed-model assembly lines with restoration to customer orders. Working paper Toth P, Vigo D (2014) Vehicle routing—problems, methods, and applications, 2nd edn., SIAM Turan B, Minner S, Hartl RF (2017) A VNS approach to multi-location inventory redistribution with vehicle routing. Comput Oper Res 78(2):526–536 Washington SP, Karlaftis MG, Mannering FL (2011) Statistical and econometric methods for transportation data analysis, 2nd edn. CRC Press, Boca Raton www.inboundlogistics.com Yao M, Minner S (2017) Review of multi-supplier inventory models in supply chain management: an update. Working paper

Chapter 12

Manufacturing Planning and Control Systems Henk Zijm and Alberto Regattieri

Abstract In this chapter, we discuss essentials of the best-known manufacturing planning and control systems. Each of these systems has its merits but each one also requires a number of conditions to be fulfilled in order to perform near-optimally, often without being explicit about these conditions. The focus of this chapter will be on discrete manufacturing planning and control with limited attention to process industries. We begin the discussion at a basic level with the most elementary result of efficiency-driven production, the Economic Production Quantity, and an extension to non-stationary deterministic demand. Next, we continue with an introduction to Materials Requirements Planning (MRP) and Manufacturing Resources Planning (MRP II), followed by a discussion of capacity oriented Hierarchical Production Planning (HPP). On a more advanced level, we introduce an entirely different approach based on the adoption of the Just-in-Time (JIT) and Lean Manufacturing (LM) philosophies. JIT and LM are more than just other planning models; they propose an entirely different approach to organizing manufacturing and assembly processes. A case study on the Toyota production system helps to understand key concepts of Lean Manufacturing. We continue with the concepts of Workload Control and the Theory of Constraints, which can be seen as means to keep internal lead times stable and hence predictable. Finally, we provide a glimpse on state-of -the-art and future developments, with a focus on digital and cloud manufacturing.

H. Zijm (B) Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, Overijssel, The Netherlands e-mail: [email protected] A. Regattieri Department of Industrial Engineering, University of Bologna, Bologna, Italy e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_12

251

252

H. Zijm and A. Regattieri

12.1 Production Under Deterministic Demand (Basic) Mass production in capital-intensive industries typically involves significant set-up costs at the start of a production run, or when changing from one product type to another—which was an important argument for Henry Ford to initially abandon any product diversity, cf. Chap. 5. In addition, mass production is usually production to stock, and these stocked products represent substantial amounts of capital invested, hence restricting liquidity of the firm, while increasing direct costs to maintain these stocks (warehousing, materials handling, insurance and interest costs). High set-up costs favor long production runs to spread the costs among as many products as possible. A new production run initially boosts product inventory and the longer the run, the higher the resulting stock. Hence, it seemed natural to seek a balance between production set-up and inventory costs, leading to the famous HARRIS/WILSON formula, also known as the Economic Order Quantity (EOQ) or the Economic Production Quantity (EPQ) formula. Because it is the most basic notion in production management, we present its derivation. See also Fig. 12.1, which displays the cyclical stock behavior when producing lots of size Q under constant demand. Define D annual demand or usage of a product per year (assumed to be deterministic and constant over time) P production rate (also expressed as maximum production capacity per year) h inventory holding costs of finished products, per product, per year K set-up costs of a single production run c marginal production costs per product Q quantity produced in a single production run From Fig. 12.1 the overall annual production costs T C(Q) as a function of the production lot size is deduced as:   D D 1 , T C(Q)  cD + K + h Q 1 − Q 2 P

Fig. 12.1 A cyclical behavior of stocks as a function of periodic production runs, under constant demand

12 Manufacturing Planning and Control Systems

253

which is minimized by setting the first derivative of T C(Q) equal to zero, leading to the EPQ formula  2K D ∗   Q  h 1 − DP Note that in case of instantaneous production (ordering), i.e., P  ∞, the EPQ formula reduces to the well-known Economic Order Quantity (EOQ). Also, note that the marginal production costs c do not play any role (they are incurred regardless of the period in which production occurs). The EPQ formula is used until today in many software packages, despite the fact that it holds only under rather severe restrictions, such as constant demand, no product quality deficiencies, and materials that are always available at the start of a production run. If demand is stochastic but stationary, i.e., with constant mean and standard deviation σ of the demand per cycle of length QD , often an additional safety stock of size kσ is added to the maximum stock level (with k being the safety factor), such that production is started as soon as the inventory level drops below  the level kσ and is continued until the level kσ + Q 1 − DP is reached. Although an approximation, this approach serves reasonably well as long as demand is unknown but can be modeled as a stationary stochastic process. Note however that a stock-out situation may occur when k is not high enough, or under rather erratic demand. When stock-out costs are explicitly known, an optimal value of the safety factor k is often found by calculating an optimal balance between holding and stock-out costs, but often k is set by the user as a function of a desired service level. The EPQ has inspired many authors to develop extensions, including discount, finite capacity and multi-product models, in both a deterministic and stochastic context, but still mainly based on the Harris/Wilson approach. An immediate extension of the situation with constant demand is the case where demand is known (hence deterministic) for a number of periods in the future, but may vary from period to period. Let D(n) be the demand for period n, n  1, . . . , N . For simplicity, we assume that the length of a production run never exceeds one period and that demand in any period can be satisfied from production in that period or from production in earlier periods. If products produced in one period are intended to satisfy a future period’s demand, inventory-holding costs of h per product per period are incurred. Let, as before, K denote the set-up costs associated with the start of a production run. Once again, we seek to balance inventory-holding costs against set-up costs. Now, some reflection on the problem yields the following properties: 1. If a production run is started, then the production run should cover demand for an integer number of periods. For example, at the start of period 1, the production run should equal D(1), or D(1) + D(2), etc. In particular, demand of the last period covered is not just partially, but fully included. 2. There is an upper limit to the number of periods m that a demand for any period j, say, is planned in advance. More specifically, we have D( j)mh ≤ K , hence

254

H. Zijm and A. Regattieri

m≤

K D( j)h

From property 1, we immediately deduce that a production run is started only if the inventory has dropped to zero. In any period, demand is satisfied by either using the goods in inventory or from a new production run. It is never optimal to satisfy demand using both options. For this problem, an exact algorithm based on Dynamic Programming exists, known as the Wagner-Whitin algorithm. The algorithm can work either forward or backward in time, but assumes that after period N the inventory should be equal to zero (or equal to a fixed given amount). In practice, however, production planners are often faced with a rolling horizon planning problem, that is, during the process, demand for future periods (later than N ) becomes known. As we will see later, deterministic demand is not unrealistic, in particular for situations in which the manufacturing of parts are to be used at a later stage in a final product assembly. In such a case, we have dependent demand as will be discussed in more detail in the next section on Materials Requirements Planning (MRP). However, fixing the inventory after N periods makes the Wagner-Whitin procedure somewhat questionable because new information on future period demand might alter the entire structure of production runs set up earlier, which is highly undesired. Therefore, we prefer to apply a heuristic procedure that has proven to yield close-to-optimal results but does not suffer from the above-mentioned drawback in a rolling horizon planning procedure. That approximate procedure is known as the Silver-Meal heuristic and works as follows: If the inventory at the beginning of some period is equal to zero then a production run should be started. To that end, calculate T (n) as follows:  K + nj1 ( j − 1)D( j)h n  1, 2, 3, . . . T (n)  n Hence, T (n) denotes the average cost per period if the production run covers demand for n periods ahead. Now, let n 0 be the first integer for which T (n 0 ) < T (n 0 + 1). Then the production run is set to cover demand for n 0 periods and the next decision will be made when that production quantity is entirely depleted (hence n 0 periods in the future). As mentioned, the Silver-Meal heuristic works perfectly in a situation with dependent, deterministic but nonstationary demand, in a rolling horizon context. If demand is stochastic, it may again serve as an approximation, but in that case a new production run occurs as soon as the inventory level drops below the safety stock level. There exist many generalizations of both procedures outlined above, but essentially they all balance set-up against inventory holding costs. The procedures are often used implicitly or explicitly in dependent demand situations to be discussed in

12 Manufacturing Planning and Control Systems

255

the next section. However, what remains is that high set-up costs inevitably lead to higher stock levels. If that is unwanted, a sound approach is to develop both technical and organizational measures to reduce these set-up costs. That is the rationale behind the Just-in-Time and Lean Manufacturing philosophy do be discussed in Sect. 12.4.

12.2 Manufacturing Resources Planning and Hierarchical Production Planning (Basic) The EPQ rule and the Silver-Meal heuristic have been developed for the case of an isolated, single stage production process, without accounting for implications upstream (materials availability) and effects downstream (use of products in subsequent production processes). In this section, we therefore turn our attention to more elaborate planning systems, of which the first one, Manufacturing Resources Planning, has gained wide acceptance in industries across the globe. Initially, the system was designed primarily for discrete manufacturing environments and was entirely materials oriented, while later many additional business functions were integrated. Hierarchical Production Planning on the other hand is primarily capacity oriented and indeed in the first place designed for process industries, although a number of features are also applicable in discrete manufacturing systems.

12.2.1 Manufacturing Resources Planning In discrete manufacturing, almost all products are built up from subassemblies that are composed from parts that are put together from assemblies, sub-assemblies, components, down to raw materials. These materials are often delivered by external partners, while also the fabrication of the various parts and components generally requires specialized equipment and operators. Clearly, the numbers of parts and components to be produced in any period are directly deduced from the number of end products for which they are needed. This is the notion of dependent demand which forms the basis of Materials Requirements Planning (MRP I), introduced in the late sixties of the 20th century, and its follow-up Manufacturing Resources Planning (MRP II). Key elements in an MRP system are the Master Production Schedule (MPS), the Bill of Materials (BOM), the inventory levels of materials, and the offset lead times, initially often expressed in weeks, and corresponding batch sizes. Figure 12.2 presents an example of the Bill of Materials of an abstract product A. Note that some components (M, P) occur more than once in the product structure, even at different levels. Next to the component indicators, the numbers of components needed in the parent part or component are displayed, e.g., product A is composed from one subassembly B and D, and two subassemblies C and E. The numbers above the

256

H. Zijm and A. Regattieri

Fig. 12.2 Bill of materials

component indicators are the offset lead times, expressed in weeks, that are reserved to produce these components. Off-set lead times are typically long to cover various effects that may cause delays, such as economic production batches, safety time to make up for quality defects, machine set-up times, limited production capacity or waiting times in case machines are used for multiple parts produced subsequently, etc. Note that, due to considerations of efficiency, lots are produced at levels larger than immediately needed (following the EPQ or the Silver-Meal logic) at each level of the BOM. As a result, inventories of parts and components at each level may arise and be depleted regularly. A typical MRP run starts with a request to produce, for instance, k products of type A in week 10. Backward calculations indicate that at the start of this production run, 2k items of subassembly C need to be available. If stock of subassembly C is insufficient, then a production run of C has to be started one week earlier, in week 9. Continuing in this way for all subassemblies, parts and components, and taking into consideration off-set lead times, we find that in case of insufficient stock at all intermediate levels, production of components O for instance needs to start 6 weeks in advance (use longest path calculations), i.e., in week 4, in order to allow production of final product A to start in week 10. Although the basic MRP logic is simple and easily explained to any manager, it immediately also reveals a number of inherent weaknesses. Crucial is that the production of end-items, based on external demand, should be frozen long in advance in order to allow all necessary components and parts to be available. The length of the frozen period depends on the offset lead times which themselves may be long already for reasons explained above. If production quantities of end-items are altered within the frozen period, crash actions are needed to still match external demand in time, or alternatively safety may be built in by either extending offset lead times further or allowing for safety stock at intermediate levels. The apparent simplicity of the MRP logic makes the procedure rigid at the same time; it simply lacks the flexibility to cope with unforeseen changes at short notice. As a result, operators often start

12 Manufacturing Planning and Control Systems

257

interventions, which, if not properly recorded, ultimately ruin MRP system’s data integrity. To reduce these risks, they often increase inventory levels and hence stock costs, in this way further compromising the system’s transparency. Nevertheless, the introduction of MRP meant a big step forward compared to classical production control methods, an improvement which is entirely due to the computing power becoming available in the sixties of the preceding century. The exponential growth of the latter now enables the daily or even continuous execution of MRP runs. Still, the basic logic remains unaltered; the use of fixed offset lead times and a rigid BOM are essential elements. Faced with today’s product diversity and in particular, the phenomenon of personalized products, the number of end-items to be produced is virtually unlimited. That brings us back to the third basic ingredient of MRP: the Master Production Schedule (MPS). It is important to realize that the items planned at an MPS level in a Bill of Materials are not necessarily the products sold to external customers (being either consumers or other industrial firms). In case of mass production (e.g., household appliances, consumer electronics) MPS items are indeed mostly sales items. However, in many cases products are personalized depending on ultimate customer demand, a phenomenon known as mass customization. Adapting a product to personal needs is possible in case of a high product modularity in which case a product is built up from a limited number of basic modules (corresponding to various options a customer may or may not select, as in the automotive and consumer electronics industry). Some fashion companies allow for the production of clothes (e.g., shirts) with customer-selected prints (or even prints electronically supplied by the customer). Such a customization is only possibly in fast, highly flexible, production systems, such as the robotized assembly lines in the automotive industry, or fully automated printing lines in case of the fashion example. Nevertheless, the basic components needed to start the customization process should already be available, and therefore these basic components are generally the MPS items used in an MRP system. The notion of MPS-items is closely linked to that of the customer order decoupling point (CODP), as discussed in Chap. 5. Recall that the CODP separates upstream production-to-stock from downstream production-to-order. For Make- and Assemble-to-Stock (MATS) systems, the MPS is generally defined at the level of final products if the product diversity is relatively low. If diversity of end-items is high but most products are built from a relatively small number of generic product types, the MPS is at the level of these generic products. The same holds for Make-toStock/Assemble-to-Order (MTS/ATO) systems where the final assembly is based on customer orders but the basic product types are produced to stock; also in such a case, the MPS is defined in terms of these basic product types. For Make-to-Order) MTO systems, in which also at least some parts are customer specific, the Bill of Materials only holds for the standard parts but typically in such a case the order manufacturing is structured as a project. The same holds for Engineer-to-Order (ETO) systems. So far, note that MRP does not check whether all planned processes can be executed within the available capacity profiles, let alone that it automatically shifts production orders forward or backward in time, to enable an optimal match between required and available resources. This lack of capacity considerations was recognized

258

H. Zijm and A. Regattieri

in the early days of MRP as an important shortcoming, leading to enhancements such as closed loop MRP and in particular Manufacturing Resources Planning (MRP II). The latter is a more elaborate functional framework for planning and control of manufacturing systems, with Material Requirements Planning still as the engine driving the lower level production schedules. The additional features concern an early specification of capacity requirements needed and, based upon that, capacity loading at both an aggregate and a more detailed level. At a low level finally, MRP II allows for the inclusion of Shop Floor Control and Vendor Control systems. Enterprise Resource Planning (ERP) Systems, based on MRP II systems, went a step further, integrating other business functions such as financial accounting and manpower planning, without adding any intelligent planning. Only when so-called Advanced Planning Systems (APS) entered the market, these deficiencies were partly addressed. We discuss APS in Chap. 19.

12.2.2 Hierarchical Production Planning As mentioned above, Manufacturing Resources Planning systems were in the first place designed for discrete manufacturing organisations and hence oriented towards the management of the materials flow. Hierarchical Production Planning (HPP) does basically the opposite. MRP has to specify in an early stage which products and parts have to be made in any period and then determines the consequences in terms of resource profiles. HPP, in contrast, determines the product groups for which capacity must be reserved, at some later point in time. It disaggregates these capacity reservations to time slots reserved for particular product families within each group, and finally determines how much time in each slot should be spent to the production of particular items within each family. The strong capacity orientation of HPP makes it particularly suitable for the batch processing or process industries where the material complexity is often lower than in discrete manufacturing, while capital assets (equipment) may represent substantial investments and therefore are to be highly utilized. In addition, product structures may differ fundamentally between discrete manufacturing systems and production in the process industries. While in discrete manufacturing we often find a convergent product structure (many items built into one final product, as specified by the BOM), divergent product structures often occur in the process industries. Typically, a single resource (e.g., crude oil) is processed into a number of alternate products, sometimes even simultaneously. HPP is able to deal with set-up times when considering capacities at the family planning level but, similar to MRP, is not designed to handle uncertainty properly. It is left to the planner to set safety stock levels, in particular, when production decisions at the item level are based upon runout times. The initial versions of HPP did not consider lead times at all, but these were necessarily included when a two-stage version of the system was developed. This extension however immediately showed the complexities that arise in multi-stage systems. Several attempts have been made to integrate Hierarchical Production Planning

12 Manufacturing Planning and Control Systems

259

and Manufacturing Resources Planning, with limited success. The basic drawbacks of HPP systems are the complexities arising in multi-stage systems and the fact that uncertainty at the various levels is not incorporated systematically. Nevertheless, the hierarchical nature represented in HPP is recognizable in many manufacturing systems, both discrete and continuous. In large process industries, we often find elaborate implementations of HPP-type of algorithms but the system never gained the same broad recognition as the simpler ERP systems. For HPP, an entire set of mathematical programming formulations has been designed to cover each of the three levels (type, family, and item), see Hax and Candea (1984). In Chap. 19, we show an example of a hierarchical two-stage manufacturing system where we outline the algorithms in detail.

12.3 Case Study: The Toyota Production System The Toyota Motor Company was founded in 1937 in Japan by the Toyoda family that had made a name as entrepreneur in the textile machinery business. Since then, up to 1950, the company had built 2,685 automobiles in total, next to almost 130,000 trucks, primarily for the military. In contrast, the Ford Motor Company in Detroit, USA, assembled 700 automobiles per day, so the family decided to study Ford’s mass production in detail to learn how to improve. They however quickly discovered that American manufacturing methods did not necessarily fit the Japanese circumstances. The Japanese domestic market was not well developed and in addition quite diverse, asking for a wide range of cars in relatively small quantities. Large batch production, as practiced in Detroit, could not be the answer. Moreover, one of Toyota’s engineers, Taiichi Ohno, wondered if large batch manufacturing of car bodies was necessary at all. In those early days, car bodies were made of steel, consisting or more than 300 different sheet metal parts that were pressed on a limited number of steel presses. Ohno quickly observed that the change of dies at these presses, needed to switch to another part, was a cumbersome task, requiring some four to eight hours change time (including test runs with the newly installed dies). But his conclusion was diametrically opposite to those of the American engineers: instead of accepting these change times as a given, and hence producing large batches for reasons of efficiency, he wondered whether it was possible to develop a simpler die changing technique, to be executed by the operators themselves instead of by specialist workers. He was remarkably successful, eventually reducing the change time till only 3 min. Once having achieved that, two further advantages came almost for free. In the first place: small changeover times and hence small batches led to much lower inventories between parts manufacturing and car body assembly. Second: part quality problems were discovered much earlier, leading to cause detection and adjustment

260

H. Zijm and A. Regattieri

of machine parameters, thereby preventing the production of large numbers of bad quality parts. Why did the many stops that initially resulted from bad quality sheet metal detection not discourage the workers? Basically because, after an initial crisis during which some 25% of all employees were fired, Toyota guaranteed a lifetime employment for all remaining workers, as well as a pay level based on seniority. In return, workers were asked to be flexibly employable, for which they received additional training. The bonus system based on the company’s profits indeed were a severe help in motivating employees to continuously improving the system they were part of, starting with jointly working to remove the cause of defective parts. Multi-skilled workers also proved to be an adequate answer to the strictly functionally structured mass production factories. The latter reflected the principle of labor division, in which also repair work was strictly separated from manufacturing and assembly. Ohno concluded that basically any repair work afterwards does not add value and that it was far better to discover them as soon as possible. Therefore, workers were encouraged to stop the line as soon as a problem was detected, either in a preceding assembly phase or with a supplied part. Subsequently, workers grouped together to find the cause of the problem and to work on solutions to prevent repetition. That marked the start of what became known as continuous improvement cycles. Periodically, teams were set together to review their entire process and to suggest improvements, to speed it up, to improve quality or to remove any non-value added activity. Quality improvement circles significantly contributed to what gradually became known as the lean Toyota manufacturing system. Observing the success of the set of work procedures at the shop floor, Ohno worked out a similar set of principles for Toyota’s incoming supply chains, in the relation with external suppliers. Again, people were made responsible for their own work. Design of parts was not just done by the Toyota engineers but instead worked out jointly with first tier suppliers’ engineers. They discouraged the practice of securing bids from multiple suppliers and selecting the supplier with the lowest bid. Instead, the focus was on selecting one supplier and working with them to develop high-quality components at reduced costs. Ohno was also fascinated by the stock replenishment systems that he had observed in American supermarkets. All products were on the shelf, in small quantities. Orders to replenish the shelf stock of an item were based on the consumption (sales) of that item. When transferred to a manufacturing company and supply chain, the basic idea is to produce or deliver parts only when stock of that part in a subsequent stage has almost depleted. That is the essence of the now famous Just-in-Time system, which made the Toyota production system an example studied by many manufacturers across the globe.

12 Manufacturing Planning and Control Systems

261

Fig. 12.3 Kanban production control system

12.4 Just-in-Time, Kanban and Lean Production (Advanced) The flood of inexpensive and reliable products that reached Europe and the United States after the two oil crises in the mid- and late-seventies of the 20th century, forced manufacturing leaders to study the causes of the undeniable successes of Japanese manufacturing. What they discovered was a set of procedures that soon became known as the Just in Time (JIT) system. What JIT clearly distinguishes from MRP systems is that it does not in the first place rely on computerized planning procedures, but, on the contrary, on organizational changes at the shop floor and a basic principle that can be summarized as: deliver parts or materials to a workstation only when they are needed. Systems following this principle are often called pull systems. Unfortunately, as with MRP systems, many managers were initially misled by the apparent simplicity of the underlying principle of JIT. This principle can best be explained by studying one particular implementation: a Kanban Production Control System (Kanban is the Japanese word for card), see Fig. 12.3. Consider products that are manufactured by means of a set of workstations that are visited in a predetermined sequence. Only a few parts or components of each particular product running over the line are available in front of each workstation on that line. A card (Kanban) is attached to each part or component. Once a workstation starts a process step for a product, it picks the required components or parts from its input stock. The cards attached to these parts are removed and sent to the preceding workstation, triggering production of similar parts at that workstation to replenish the input stock of its successor. The preceding workstation in turn picks the relevant components to produce the part from its own input stock, removes their Kanbans which are sent backwards. Hence, in this way, a final demand generates a sequence of replenishment orders all the way backwards through a sequence of workstations, and finally to materials procurement. If the in-process-inventories (work-in-process or WIP) are low or almost zero, one may indeed envision delivery on demand or, alternatively, of Just in Time production. Although the basic procedure outlined above is extremely simple, the system only works provided several basic conditions are fulfilled. These conditions are best summarized as: 1. The ability to realize superb quality of both products and processes in every step of the production process. 2. The avoidance of any activity that does not add value.

262

H. Zijm and A. Regattieri

Indeed, the Just in Time system is closely related to Total Quality Control, but a key difference with western production systems is the way that is achieved, i.e., by designing a production system capable of delivering high quality, zero defects products. Managing the production using Kanbans, or more general, pull strategies, requires a manufacturing system that is flexible with respect to the product mix, prompt to change lots with small setup times and working with an inbound supply chain able to supply materials in small lots and frequently. A way to move in this direction is the Lean Production concept application. In order to understand the basic principles of Lean Production, let us revisit the dominant characteristics of western production, which in the sixties of the 20th century was still grounded in mass production, based on division of labor and efficiency principles. Division of labor, as outlined in Chaps. 3 and 5, favors specialization, i.e., breaking down jobs into relatively simple repetitive tasks enabling workers to follow the learning curve and become highly efficient, but only in performing these designated tasks. With efficiency or productivity being the leading performance indicator, workers are keen to meet their quota, independent of the immediate need of their products in subsequent production stages (for this reason such production systems are also called push systems). As a result, large amounts of stocks between subsequent production stages were not uncommon, or even recommended; if a part needed for a subsequent stage did not fit due to quality problems, it was thrown away and a next part was taken, hence buffer stocks might help to achieve productivity targets in spite of defective parts or components. If a machine were needed to produce several distinct parts that are assembled together to compose a final product, typically changeovers have to take place between production runs of different parts, often at the cost of significant changeover times to replace, for instance, special tools at the machine, to perform a trial run, possibly readjust the tools, etc., until the produced new parts conform to specifications. Large changeover times however require large batches to be produced (recall the EPQ formula in the beginning of this chapter). As a result, large batches of several parts were stocked until eventually used in a final assembly. Large piles of stock mean long lead times, as a result of what in queueing theory is called Little’s formula. Large stocks need a considerably long time for depletion. Often, the amount of stock is expressed in time in western factories, for example, two weeks of stock of a specific part indicates sufficient stock to fill two weeks of work. However, upon use of these parts in a subsequent phase, if quality problems are detected, one may wonder what is the value of all these parts, produced efficiently or not. Even worse, if a part type does not meet its quality standards in an assembly phase, the entire assembly process comes to a standstill, rendering all other part type stocks temporarily obsolete as well. Problems might be less severe in case defective parts can be repaired, hence special repair shops were installed at many factories where specialists took up the task to rework such parts, or still throw them away in case repair turns out to be impossible. But even in these cases, no direct feedback was given to the worker or machine operator at the stage where the problem occurred. In addition, recall that

12 Manufacturing Planning and Control Systems

263

finding quality problems at a later stage and repairing products is not a value-added activity. The above sketched system might still work under a mass production paradigm, i.e., producing a limited number of final products in large quantities, although it is clear that the cost of the system is high, not only due to the large amounts of capital tied up in inventory but also the fact that not seldom a (sometimes considerable) percentage of parts stocked appeared to be useless due to quality problems, or had to be reworked at considerable additional effort. Gradually, it was learned that efficiency and effectiveness are not the same thing, i.e., that working efficiently is a necessary but not a sufficient condition to reach desired goals, in this case, producing desired quantities of good quality products at minimal cost. When studying western production methods, Japanese manufacturing engineers quickly realized that the mass-production philosophies would not work in building a modern manufacturing industry in their country, especially when production volumes are small. More important, they concluded that, despite efficiency being the driving force in western systems, much effort of workers was in fact wasted because of product defects that in addition were not traced back to their causes, simply because it took too long before these failures were detected due to the presence of large stocks. The main idea behind almost all Lean Production techniques is to focus on, and subsequently remove, all non-value adding activities, or to eliminate waste (called Muda). The lean approach recognizes seven typical Muda: transport (moving products that are not actually required to perform the processing), inventory (all components, work in process, and finished products not being processed), motion (people or equipment moving or walking more than is required to perform the processing), waiting (waiting for the next production step, interruptions of production during shift change), overproduction (production ahead of demand), over processing (resulting from poor tool or product design creating activity), and defects (the effort involved in inspecting for and fixing defects). A set of alternate techniques are usually applied to delete or at least to reduce Muda. We do not to elaborate the lean approach in any detail here, but some important steps are briefly discussed. A first step is to remove large batches by eliminating what causes them: high changeover times. That idea became the basis of the SMED (Single Minute Exchange of Die) technology, developed by Shigeo Shingo and others who persistently sought to modify machine tools such that they could be flexibly adjusted to produce new part types (Shingo 1985). A second step is to make all employees in a production line together responsible for end-product quality, instead of just asking each individual employee to match production targets. If any worker picking a part from the entrance stock of his workplace detects that part to be unusable, he or she is authorized to stop the entire production line after which all workers help to trace the cause of the problem and to ensure similar problems do not occur in the future. The key idea is that any effort spent on a part that eventually cannot be used is a waste of effort and does not add any value. The continuous feedback loops ensure that all steps in the process eventually yield high quality parts. In addition, once the entire production process is reliable and flexible (with only small set-up times), work-inprocess inventories can be lowered to a minimum, hence reducing throughput times

264

H. Zijm and A. Regattieri

considerably and thereby making the entire process more responsive. It is important to realize that the resulting Just-in-Time production system can only be effective in a lean manufacturing environment. A third step is to introduce what has become known as quality circles or continuous improvement circles in which all workers involved in a production process, exchange ideas to further improve the process, again fed by the idea that any action should add value to the end-product to be produced. In fact, such an approach helps all workers/operators take ownership of the production system (process and products) they are part of. Indeed, one of the key characteristics of lean production is the high worker motivation stemming from the fact that all workers are responsible for the system as a whole, not for a small part of it. This is in sharp contrast to the division of labor philosophy dominating production in Europe, the United States, and Canada. The extension of the basics of Lean Manufacturing to external part suppliers and thereby to the entire supply chain, as well as to intensive customer and dealer relationships is a natural next topic but such a discussion is beyond the scope of this chapter. Another attempt, based on the same philosophy of joint responsibility and short lead times is that of Cellular Manufacturing, in which a group of workers in one production cell is entirely responsible for a small group of end-product types. Cellular Manufacturing may be viewed as the opposite of the classical, functionally arranged, job shop facilities (cf. Chap. 5); the idea however gained limited response, but in last few years is receiving renewed attention, partly in response to the increased awareness of the importance of a motivated and skilled work force as a key production factor.

12.5 Workload Control and Theory of Constraints (Advanced) In order to keep stocks low and internal lead times small, the number of products running on a single production line has to be constrained in Just-in-Time production systems to prevent items of many different parts waiting too long before being used between workstations and to detect quality problems as early as possible. The key success of many lean production systems has been the relentless removal of any nonvalue adding activity in order to arrive at stable and responsive production systems. Nevertheless, also Just-in-Time systems cannot easily handle large fluctuations in total demand or abrupt shifts in demand between products that cannot be built with the same tools. Indeed, the system works best after smoothing and subsequently freezing monthly production schedules. The achievement of relatively short and stable internal manufacturing lead times in turn helps to establish highly reliable Master Production Schedules. However, product demand develops independently. Current manufacturing systems are increasingly characterized by a high product diversity and in addition relatively short product life cycles, which have brought authors to seek for production

12 Manufacturing Planning and Control Systems

265

management methods that maintain the advantages of lean production while being able to produce a large product variety with limited means. In particular, the question is how to reach short and stable internal lead times, given existing capacity constraints, and hence without the need to include uncertainties induced by multiple sources (quality defects, process failures) as was the case in initial MRP systems. It is this challenge that has led many authors to promote workload control as a guiding principle in Manufacturing Planning and Control. Basically, they suggest releasing production orders to a work cell or job shop (consisting of multiple work stations), only when the work load already present in the cell drops below a certain level. In this sense, a Kanban system is a highly specific implementation of workload control, on the level of individual workstations. Variants of workload control may either put (non-equal) constraints on the number of parts per type in the systems or one overall constraint on the total number of parts in the system, or define family-based constraints; such variants often go hand-in-hand with priority mechanisms at individual workstations. A rather unorthodox way to guide managers through the principles of workload control was used by Eli Goldratt who developed the Theory of Constraints (TOC) and decided to explain it in a novel. A full explanation of the Theory of Constraints is beyond the scope of this chapter but the essential idea is similar to what has been advocated in lean manufacturing, i.e., to not accept constraints as a given but attempting to remove them, and, if not possible, to cope with them in a clever way. The Optimized Production Technology (OPT), as explained in Goldratt and Cox (1984) applies the TOC to the shop floor in a factory and is based on the drum-buffer-rope principle that follows a three-step approach: 1. Identify which workstation is the bottleneck. This workstation determines the pace of the entire manufacturing system (the drum). 2. Make sure that the bottleneck workstation never runs idle due to a lack of input parts (for example caused by upstream problems). Therefore, define a time buffer in front of the bottleneck machine that is always filled (the buffer). 3. Define shop floor release and shipping schedules that are closely tied to the pace of the bottleneck workstation (the rope). It is easy to see that the drum-buffer-rope principle defines a particular implementation of a workload control system. However, what is not explained is how to determine the bottleneck workstation and moreover, how to act in case of shifting bottlenecks, depending on the actual mix of parts and products present at the shop floor. More generally, it is not clear how to quantitatively determine the relation between workload and lead time in a mixed model production system. This holds even more in a MTO production systems in which customer order delivery dates are often set without taking into consideration actual or future workloads, if known at all. One way to determine the relationship between load and internal shop manufacturing lead times is by modeling the shop as a multiclass Closed Queueing Network; see Chap. 19 for an introduction to the topic. In the same chapter, we present a shop floor scheduling algorithm, the Shifting Bottleneck procedure, that describes in detail how to schedule jobs at a (temporary) bottleneck as well as on adjacent workstations.

266

H. Zijm and A. Regattieri

Still, much work remains to be done to further develop and refine these models, and to extend them to multi-stage production systems. There is an urgent need for integrating workload control and resource availability planning on a higher level, or even supporting order acceptance by sophisticated load-based procedures. We will return to this topic in Chap. 19 where we discuss Advanced Planning Systems.

12.6 Digital and Cloud Manufacturing (State-of -the-Art) Digital Manufacturing has been around for quite a while, albeit under different names of which Computer Integrated Manufacturing and Engineering is a widely accepted one, cf. Rembold et al. (1993). In general, we may refer to three different dimensions in automation: manufacturing and logistics processes, manufacturing planning and control, and manufacturing design and engineering. Let us take a brief look at each dimension: • Manufacturing and Logistics Processes: automation of manufacturing processes started with the introduction of Direct Numerical Control (DNC) in which coding of machine instructions took place directly at a machine that was able to execute, e.g., metal cutting (milling, drilling, lathing) operations. In Computer Numerical Control (CNC) such codes may be generated automatically from a (digital) product database, together with machine and cutting tool data. Other examples are found in electronics manufacturing, e.g., machines that automatically place components on printed circuit boards. The use of robotics in assembly processes constitute another example of automation in production, and the same holds for the application of automated storage and retrieval systems (AS/RS) in warehouses, and of Automated Guided Vehicles (AGV’s) and accumulating conveyor systems in both manufacturing and distribution systems. Gradually, these systems were guided by advanced product information systems (ranging from barcode to RFID) but still they operate in their local environment, based upon information that is attached to the products. • Manufacturing Planning and Control: the use of the computer to perform routine tasks faster than any human planner was recognized already in the late sixties of the 20th century, and marked the start of Manufacturing Resources Planning (MRP). There was hardly any intelligence put into these systems. They merely served to provide quick overviews of tasks to be executed, for example, to order new materials from suppliers. This situation eventually changed with the development of so-called Advanced Planning and Scheduling systems, based on methods from Operations Research (OR) and Artificial Intelligence (AI). • Manufacturing design and engineering: simultaneously with the developments in hardware automation, the computer was discovered as a tool to support engineering processes, and in fact to evaluate many alternative designs quickly. Feature based design (cf. Chap. 5) is greatly facilitated by computer support and the same holds for the translation of a digital design into process and machine instructions. That

12 Manufacturing Planning and Control Systems

267

was the introduction of Computer Aided Design (CAD) and Computer Aided Process Planning (CAPP). Still, the computer served in the first place to execute and speed up routine activities, without the application of intelligent algorithms. Currently, digital manufacturing is sometimes used as a synonym for smart manufacturing, indicating next to straight automation of existing processes more sophisticated techniques are being implemented, such as the Advanced Planning systems discussed above. An important development is the application of machine learning techniques, in which machines (or systems) automatically deduce algorithms based on experienced practices (case-based reasoning) or on a narrow description of constraints and potential options (rule-based reasoning). The design of expert systems is often based on such AI applications. When extended to industry at large, often the term Smart Industry or Industry 4.0 (in Germany) is used as a common designation for the (artificial) intelligence added to current manufacturing systems. Next, we turn to cloud computing, which is a key aspect of the digital society in which we live today. The way the computer has changed our lives is nothing less than a revolution. Starting with mainframe and mini-computers, quickly computer power came within reach of any human being, via the desktop, laptop computer, and now hand-held devices. The next step was the integration of Information Technology and Communication, now designated as ICT, which enabled computers to interact with each other, leading to the development of the Internet and the World Wide Web, with its standard addresses and message transfer protocols. In fact, the Internet is an early example of what is now known as Infrastructure as a Service (IaaS), a standardized infrastructure that allows free communication, not only between persons, but also between machines and more and more between any devices that need to signal their status to start a response action from another device. That is the Internet of Things, which quickly has grown into a fundamental tool to allow for fast decision making without human intervention, based upon smart algorithms. One further development is that of Software as a Service (SaaS). Initially, most software packages were to be purchased and used on local machines owned by the user. In fact, there is no need to own the software provided one can make use of it, via a system of subscription and payment of a periodic fee. The development of a fast and reliable communication infrastructure obviously has greatly enabled the development of SaaS, running from professional (industry oriented) applications to home services such as streaming audio and video. Naturally, such applications bring about the processing and transfer of vast amounts of data, which in turn has fruitfully influenced developments in fields such as data and process mining, and data analytics (often termed the “big data” revolution). Still one further step is the decision not to own computer power anymore but to rely on centrally available power, and to execute major computing tasks on central platforms. With that, we have reached the full potential of cloud computing. Indeed, a number of computer industries currently face the change from selling hardware and software to running major datacenters to which clients may subscribe and use both its computing power and software services, while only the results are transferred to the client’s base. That does not mean computer power is absent at the clients’ base;

268

H. Zijm and A. Regattieri

when discussing the emerge of the Internet of Things we already noted that a large variety of devices, from household appliances to consumer electronics to cars and industrial machinery, currently is equipped with integrated hardware and software, able to execute the computing and communication tasks needed to interpret and process the results of the externally executed tasks. For an elaborate treatment of cloud computing and Internet of Things, see Chap. 8 of this volume. Cloud manufacturing is sometimes described as the manufacturing version of cloud computing. It refers to the transformation of manufacturing resources and manufacturing capabilities into manufacturing services, to be managed and operated externally. Cloud manufacturing is usually discerned into two categories: • The first category can be seen as the equivalent of SaaS and concerns the deployment of ERP software or the use of CAD or CAPP applications that are supplied as a (remote) service in the cloud. • The second category is much broader and in fact allows for the execution of physical manufacturing processes, as well as tasks related to the design and engineering of products externally (in the cloud). The latter has been occurring for quite some time, enabled by fast EDI (Electronic Data Interchange) systems but the former really means a breakthrough in the way we think about manufacturing systems. Xu (2012) has provided the following definition, which covers both categories: Cloud manufacturing can be seen as a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable manufacturing resources (e.g., manufacturing software tools, manufacturing equipment, and manufacturing capabilities) that can be rapidly provisioned and released with minimal management effort or service provider interaction.

There are a number of reasons to embrace the possibilities of cloud manufacturing. Its basis is digital manufacturing which allows for an easy exchange of product and process design data. Further, as we have seen in our discussion of MRP and HPP, the internal match between required and available resources at a company level becomes more and more difficult in a time in which short product life cycles and high customization becomes the rule rather than the exception. Cloud manufacturing therefore may help companies on their path to become agile manufacturing enterprises that are able to quickly adapt to changing market demands. Not owning all production means simply reduces risks, and companies are willing to pay for that. In this sense, cloud manufacturing may be seen as an example of what has become known as the sharing economy in which a user benefits from a functionality, without necessarily taking ownership of the platform that provides the desired function. Partly, the decision to exploit cloud manufacturing may be similar to make or buy decisions: if some manufacturing resource is only temporarily needed, it makes sense to not buy the equipment, rather rent it as needed. However, the scope of cloud manufacturing is much wider, in that one may exploit the Internet or specialized networks to search and find the most suitable partner, that may change over time as well. Currently, quite a number of companies investigate the pros and cons of additive manufacturing (3D printing) through the search of appropriate 3D printing service providers via the internet.

12 Manufacturing Planning and Control Systems

269

However, the way to maturity of cloud manufacturing is still a long one and important hurdles remain. These are related to interoperability aspects such as standard protocols and the standardization of interfaces, to the development of Quality of Service standards of both products and processes, to the need for pay-per-use billing systems and last but not least, to security issues (violation of intellectual property rights). In general, manufacturers are still reluctant, for reasons of competitiveness, to share information on product characteristics and market positions with third parties. An overview of both potentials and concerns can be found in He and Lida (2015). Cloud manufacturing is a new step beyond already long established distributed manufacturing systems but, like the latter, call for entirely new business models that support transactional arrangements between various partners that may change over time. Transaction models based upon agreed service level arrangements that enable a fair share between collaborating partners often apply game-theoretical methods, while the process to come to mutual agreements is often modeled as a multi-agent system. Multi-agent systems are discussed in Chap. 27.

12.7 Further Reading The origin of the Economic Production Quantity can be traced back to Harris (1913), while the concept was explored in depth by Wilson (1934). It gave rise to an overwhelming number of papers dealing with extensions and generalizations, see Sect. 4.9 in Silver et al. (2017) for several examples. The MRP crusade started with the book of Orlicky (1975), while MRP II was advocated by Wight (1981). A splendid overview can be found in very complete book of Vollmann et al. (1997). A fine introduction to Hierarchical Production Planning models is provided by Hax and Candea (1984) although the initial ideas were published already by Hax and Meal (1975). An extension of multi-stage systems is presented by Haas et al. (1982), of which a modified version is discussed in Chap. 20. Meal et al. (1987) present a further attempt to integrate MRP and HPP. Zijm (2000) discussed some of the deficiencies on the MRP methodology, in particular, its lack of a thorough capacity planning module. Schonberger (1982) presents a thorough overview of the essentials of what became known as the Just-In-Time production system, with the developments at the Toyota Motor Company as its most famous example. Another introduction to the topic is Monden (1998) while the SMED technique is outlined in detail, with many examples in Shingo (1985). Zijm (2000) pointed out the strict conditions under which JIT should work, such as a high system reliability. The principles of Lean Manufacturing are presented by Womack et al. (1990), geared towards the automotive industry. Some elements of the case study on the Toyota Production system are based on information in this publication. Bertrand et al. (1990) and Wiendahl (1993) have advocated the principles of workload control as a means to stabilize internal lead times; another approach known

270

H. Zijm and A. Regattieri

as CONWIP (CONstant Work In Process) has been elaborated in Hopp and Spearman (2000). The novel of Goldratt and Cox (1984) on the Theory of Constraints became a must-read in many manufacturing organizations. An early introduction to Computer Integrated Manufacturing is the book by Rembold et al. (1993). Cloud computing and IoT is extensively discussed in Chap. 8 of this volume. A highly readable introduction to cloud manufacturing is presented by Xu (2012), while He and Lida (2015) review the most significant recent developments, see also Lee et al. (2014). Collaborative design and planning methods are discussed by Wang and Yeh Ching Nee (2009).

References Bertrand JWM, Wortmann JC, Wijngaard J (1990) Production control: a structural and design oriented approach. Elsevier, Amsterdam Bitran GR, Haas EA, Hax AC (1982) Hierarchical production planning: a two stage system. Oper Res 30:232–251 Goldratt E, Cox J (1984): The Goal—a process of ongoing improvement. North-River Press Harris FW (1913) How many parts to make at once. Fact Mag Manag 10(2):135–136, 152 Hax AC, Candea D (1984) Production and inventory management. Prentice-Hall, Englewood Cliffs, NJ Hax AC, Meal HC (1975) Hierarchical integration of production planning and scheduling. In: Geisler MA (ed) Logistics. Studies in the management sciences, vol 1. Elsevier, North-Holland He W, Lida X (2015) A state-of-the-art survey of cloud manufacturing. Int J Comput Integr Manuf 28:239–250 Hopp WJ, Spearman ML (2000) Factory Physics: foundations of manufacturing management, 2nd edn. Irwin/McGraw-Hill, New York Lee J, Bagheri B, Kao H (2014) Recent advances and trends of cyber-physical systems and Big Data analytics in industrial informatics. In: IEEE International Conference on Industrial Informatics (INDIN) Meal HC, Wachter MH, Whybark DC (1987) Material requirements planning in hierarchical production planning systems. Int J Prod Res 25(7):947–956 Monden Y (1998) Toyota production system: an integrated approach to just-in-time, 3rd edn. Engineering Management Press, Norcross, GA Orlicky J (1975) Material requirements planning: the new way of life in production and inventory management. McGraw-Hill, New York Rembold U, Nnaji BO, Storr A (1993) Computer integrated manufacturing and engineering. Addison-Wesley Ltd Schonberger RJ (1982) Japanese manufacturing techniques: nine hidden lessons in simplicity. The Free Press, New York Shingo S (1985) A revolution in manufacturing—the SMED system. Productivity Press, Portland Silver EA, Pyke DF, Thomas DJ (2017) Inventory and production management in supply chains. CRC Press (Taylor and Francis), Boca Raton Vollmann TE, Berry WL, Whybark DC (1997) Manufacturing planning and control systems, 4th edn. Irwin/McGraw-Hill, New York Wang L, Yeh Ching Nee A (2009) Collaborative design and planning for digital manufacturing. Springer Publishing Company, London Wiendahl H-P (1993) Load-oriented manufacturing control. Springer, Berlin, Heidelberg, New York Wight O (1981) MRP II—Unlocking America’s productivity potential. CBI Publishing, Boston Wilson RH (1934) A scientific routine to stock control. Harv Bus Rev 13:116–128

12 Manufacturing Planning and Control Systems

271

Womack JP, Jones DT, Roos D (1990) The machine that changed the world. Maxwell Macmillan, New York Xu X (2012) From cloud computing to cloud manufacturing. Robot Comput Integr Manuf 28:75–86 Zijm WHM (2000) Towards intelligent manufacturing planning and control systems. OR Spectr 22:313–345

Chapter 13

Packaging Logistics Alberto Regattieri, Giulia Santarelli and Francesco Piana

Abstract The chapter discusses the important role of packaging as part of a company’s market and operations management. Packaging has been evolving for centuries and continues to do so, forcing consumers to change their habits. From simple protection tools to safeguarding carried items, packaging has become a marketing and communication instrument as important as contents. E-commerce and onlineretailers growth change again the packaging role during purchasing, bring it back as mere logistics and protection tool. In the first basic section, a brief history of packaging is described. Starting from the most important steps, which characterize packaging from the very beginning, in prehistory, followed by the commercial discoveries of the Middle Ages, the industrial revolution, until the present day with the newest and most ingenious packaging solutions. With a focus on marketing, design, logistics, environmental impact, and costs, a complete reference framework underlining how each of these aspects change along the supply chain steps is presented. The advanced and second section of the chapter presents case studies, models and methods for packaging costs evaluation and unit load design. Logistics and packaging are fundamental business leverages for a modern company, therefore any improvements can represent opportunities and generate profits. The chapter ends with a discussion about state-of -the-art research in packaging: the last decade showed a strong development of the e-commerce market, and the central role of packaging changed. Furthermore, the evolution of technologies, materials and Internet Communication Technologies in the last few years have improved the packaging world as well creating, for example, intelligent packaging that can communicate with consumers or interact with products, allowing the traceability of a single item within the entire supply chain. Packaging is a fundamental resource for end-customers and all actors involved in the entire supply chain.

A. Regattieri (B) · G. Santarelli · F. Piana Department of Industrial Engineering, University of Bologna, Bologna, Italy e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_13

273

274

A. Regattieri et al.

13.1 History, Modern Function and References Framework of Packaging (Basic) The earliest forms of prehistoric packaging were made from material found in nature, such as animal skins, wood and vegetables. Water was kept in containers made from wood, bamboo, shells and animal skins. Glass, which emerged in the Far East around 5,000 years B.C., was one invention destined to revolutionize life and the capacity to conserve and transport goods, especially liquids. Egyptians used this material to create jars of many kinds. In the Medieval age, barrels became the most frequently used way of preserving goods. They were used for storing different types of solids and liquids, protecting them from light, heat and moisture. They were easy to load and transport, thanks to the rolling shape, by cart or ship. Especially during the ages of discovery, barrel become precious as a container for water storage and transport. Industrial Revolution in Europe not only changed companies, working methods and society, but introduced innovations and new materials. For the first time, tin sheet jar was used for food conservation. Metals become new packaging materials. The vast range of new products made available to the consumer caused a change in lifestyle, providing consumers with greater choice and allowing trade to flourish. As is typical, many innovations are developed first for military uses and then applied to commercial uses. This is true for packaging materials as well. Military packaging had to transport materials, supplies, food, and other items under the most severe distribution and storage conditions. In response to Napoleon Bonaparte’s call for a new and an innovative idea for storage and preserve food for soldiers, Nicolas Appert invented and presented the metal tinned can in 1810. Glass also become useful as a method for preserving food. At the end of the 19th century, an American, Robert Gair had the bright idea of manufacturing a pre-cut cardboard panel, which would form a box after folding. Cardboard boxes become easier and lower cost methods for transporting and storage goods. Polyethylene was the first plastic used in packaging during 1920, after the invention of cellophane. From then on, other technical innovations led to the continued improvement in packaging logistics, entering the household life. Currently, packaging is much different from that developed thousands of years ago. Packaging lies at the very heart of modern industry, and efficient packaging is a necessity for almost every type of product, whether it is mined, grown, hunted, extracted or manufactured. It represents an essential link between the product makers and their customers, and unless performed correctly, the reputation of the product will suffer and the goodwill of the customer will be lost (Titus and Ahearn 1992). Packaging enhances and protects goods, from processing and manufacturing through handling and storage to the final consumer. Moreover, packaging helps in identifying products on the shelf, distinguishing them from their competitors. Without packaging, materials handling would be a messy, inefficient and costly exercise and modern consumer marketing would be virtually impossible.

13 Packaging Logistics

275

Fig. 13.1 Primary, secondary, tertiary, and system packaging (http://trayak.com/transport-packag ing-and-compass-lca/)

Packaging is built up as a system usually consisting of a primary, secondary, and tertiary level (Saghir 2002; Chapman et al. 2003; Davis and Song 2006; Jönson and Johnsson 2006). The primary package is usually the smallest unit of distribution or use and is the package in direct contact with goods (especially when the package contains food, a plastic bag or film can be put between the package and the contents). The secondary package is used to group primary packages together and has the purpose of visual communication. The function of the secondary package is to protect the products during the transportation to the final destination and provide information for the user. Finally, the tertiary package is used for storing and transporting products. The various levels and their interactions should be regarded as a packaging system (Saghir 2004; Olsson et al. 2004). The performance of the packaging system is affected by the performance of each level and of the interactions between them (Hellström and Saghir 2007). Figure 13.1 shows the three levels of the packaging system.

13.2 Modern Functions of Packaging Throughout history, packages have been used to contain, store and transport goods and products (Beckeman et al. 2007; Hellström and Saghir 2007). The fundamental logistics function of packaging is mainly to protect products during movement through distribution channels (Lockamy 1995; Hermansson 1999; Saghir and Jönson 2001; Dominic 2005; Klevås 2005; Hellström and Saghir 2007; Olander-Roese and Nilsson 2009; Prendergast and Leyland 2009) and, on the other hand, to protect the environment from products (Klevas 2006; Hellström and Saghir 2007). Packaging is an important competitive factor for companies to operate an efficient supply chain. It is an essential element and in many cases, without its use, product handling would be inefficient and impractical (Lockamy 1995). Packaging contributes to the success of product supply chain, enabling efficient distribution of products and reducing the environmental impact of product spoilage and waste (Verghese and Lewis 2007). According to Paine (1990), better packaging can reduce cost, increase turnover, minimise damage complaints, and reduce waste.

276

A. Regattieri et al.

Fig. 13.2 Simplified manufacturing company supply chain Table 13.1 The role of packaging for the actors of the supply chain

Actors

Role of packaging

Suppliers

Logistics and transportability Minimization costs Protection and safety Logistics Marketing Environmental Logistics and transportability

Manufacturer

Wholesaler Retailer End consumers

Marketing Environmental Marketing Protection and safety Environmental

During the last few decades, the role of packaging and its functions have grown considerably due to new and different developments of packaging materials and methods (Domnica 2010). Packaging has become a critical/versatile marketing tool for sales promotion, customer attention and brand communication and is a vehicle to provide customers with product information and promote a product through its visual appeal, to attract customer attention and to create a positive impression (Gray and Guthrie; Sara 1990; Olander-Roese and Nilsson 2009; Prendergast and Leyland 2009). Hence, packaging not only preserves the quality of the main product but also provides an opportunity to create added value to customers (Olsson and Györei 2002; Olander-Roese and Nilsson 2009) and to other actors in the packaging supply chain. Figure 13.2 underlines the considerable role packaging has within all steps along a simplified supply chain of a manufacturing company. Each actor has different main needs and requirements regarding packaging, summarised in Table 13.1. In the table below only the most influential needs and requirements for each actor of the whole supply chain are highlighted. Clearly some needs like costs minimization or items protection are purposes of all actors, but in some cases and peculiar situations some packaging requirements are more relevant than expected: i.e. cork caps in wine bottles, despite being more expensive and less environmental friendly than synthetic or metal caps, guarantee better wine organoleptic proprieties conservation (Silva et al. 2012).

13 Packaging Logistics

277

Table 13.2 A graphical representation of the packaging framework Marketing Design Logistics Environment Attraction Communication Promotion Customisation Customer satisfaction

Dimensions Volume Weight Materials Protection and preservation Resistance Convenience

Containment Unitisation Standardisation Storage and distribution Traceability Reverse logistics

Waste reduction Pollution minimisation Recycling Re-use Sustainability

Cost Efficiency Cost parameters

Packaging design is an element of the Operations Management discipline and must be integrated in the product design process, taking into account logistics, production, marketing and environmental needs (Regattieri and Santarelli 2013). Although the relevant role of packaging is recognised by all the actors along the supply chain, there are inefficiencies moving from the packaging supplier to the packer/filler to the point of sale and consumer (Dominic 2005). There is not enough coordination between parties in the study and development of the entire system of packaging: often, the packaging supplier is not involved in the activities in the market place (Dominic 2005) and important upstream information from the market place regarding the consumer is not always released on time to the supplier to produce packages that are suitable for the consumer. Dominic (2005) created a knowledge base that proved to be useful for the packaging industry, identified the network integrators and analysed the development for packaging suppliers.

13.3 Packaging Logistics Reference Framework The role of packaging as fundamental drivers in integrating the management of industry functions along the entire supply chain cannot be ignored. Then, a framework is presented for evaluating the packaging role with emphasis on a wider system. Five packaging key drivers are defined to develop an innovative reference framework: marketing, design, logistics, cost and environment, each having fundamental importance for the final resolution of the product package at minimum cost. To increase efficiency and effectiveness and reduce costs, it is necessary that the packaging key drivers act in a coordinated and collaborative way and communicate the necessary information to each other. The reference framework of packaging is presented in Table 13.2.

278

A. Regattieri et al.

First, the marketing function of packaging is analysed. Through its aesthetic function, packaging contributes to making the product more attractive and affects consumers during their purchase decision. Packaging attractiveness could be realised in several ways, e.g., colour, shape, size, all with the purpose to attract the consumer and bring him or her to buy that product. Another fundamental aspect of the marketing function of packaging is communication that allows consumers to recognise instantly products through distinctive branding and labelling. According to Olsson and Larsson (2009), the communication function of packaging is threefold: communication of necessary information, communication of promotion, and communication to consumers. Packaging must communicate all necessary information (e.g., content, expiration date, materials, etc.) regarding the product and the package to stakeholders and consumers. The second communication aspect concerns the promotion of the product. Promotion is the incentive that makes a product known and appreciated, and its main objective is to affect the consumers. Packaging must communicate the product benefits, attract consumers, and hold their attention against the visual glamour of competitive products. This means that packaging should be a differentiating element during the purchase process. The product and the package are often perceived as closely integrated and the consumer initial impression of the quality and value of a product is sometimes determined by their packaging perception (Silayoi et al. 2004; Olsson and Larsson 2009). If packaging communicates high quality, consumers assume that the product is of high quality and, vice versa, if packaging communicates low quality, consumers regard it as a low-quality product. Concept behind Internet of Things is to give to real objects an active role by means of a network link: a QR code in packaging offers to producers a virtual space to fill with information about a product otherwise impossible to write on a simple label. In situations where the label is small, such as wine bottles as shown in Fig. 13.3, this solution should be very useful: producers can write on a product-dedicated webpage many different information in different languages. Another interesting characteristic of packaging relating to the marketing function is customisation. Marketing prefers variation, specialities and different packaging sizes to be adaptable to each single customer need. Packaging customisation could be a fundamental aspect to increase customer satisfaction, which is crucial for the success of companies (Matzler et al. 1996). Achieving customer satisfaction means understanding and anticipating what users want, but do not expect, from the product in the future. The most important aspect is to delight customers and surprise them. A crucial element is to identify the product qualities that are decisive for satisfying the costumer and the features that may prevent satisfaction. Packaging design is the second key driver and considers both the physical and mechanical characteristics.

13 Packaging Logistics

279

Fig. 13.3 A refined packaging of women’s perfume (http://www.agi.it/pictures/agi/agi/2016/04/1 1/114428611-253bda52-2d2e-4c4c-93e1-044e9b6310de.jpg) and a QR code on a wine bottle label (https://s-media-cache-ak0.pinimg.com/736x/94/ec/0e/94ec0e24a604afeefd66fe3c1fd3f24c.jpg)

From the physical point of view, the shape of a package (in terms of dimension, volume and weight) is considered relevant. The increase of costs along the supply chain is due if the volume and weight of package are not designed and then made in an efficient way. Economic, but also environmental aspects require that the package should be reduced in terms of dimension, volume and weight as much as possible. Materials used for realising a package constitute another important aspect affecting the design process. The packaging designer should encourage the reduction of materials used per unit of product. Not only to save costs but also to facilitate manufacturing operations, handling, transportation, and packaging disposal. To facilitate package recycling by end-consumers, mono-material or biodegradable packaging is preferred. Mechanically a package should cover functions of protection and preservation of the product in the right conditions during both the exposition on the shelf and the transport, as well as having high resistance to vibrations and shocks during handling and distribution. The protection of the product is a fundamental function that a package should cover and concerns the protection of the product from the environment and the protection of the environment from the product. The choice of the protection degree of products depends on the economic value and fragility of the products themselves. Another fundamental characteristic affecting packaging systems is the ease of use. It can simplify the use of products by the consumer, making handling easy and user-friendly as much as possible (e.g., to open, re-close and grip) (Fig. 13.4). A framework on packaging must consider the logistics aspect too. The first important characteristic is related to the containment of products. It is primarily responsible for restraining the contents of packaging (Lockamy 1995). Another interesting aspect is the unitisation function of packaging, which permits primary packages to be unitised into secondary packages and secondary packages to be unitised into tertiary

280

A. Regattieri et al.

Fig. 13.4 An example of super-protective packaging for bottles (http://www.grifal.it/wp-content/ uploads/2015/03/UPS_MG_7135.web_-460x295.jpg) and easy-to-use packaging for candy (http:// www.break.it/wp-content/uploads/2015/05/01slide_3_mark.jpg)

packages for the efficient transport of packed products. Unitisation facilitates the optimisation of material handling activities by reducing the number of individual packages or loads that require handling. The result of unitisation is called Unit Load. To optimise the unitisation process, it may be necessary to adopt a standardisation. Packaging standardisation (i.e., the use of a limited number of different sizes of packages for the transportation and handling of products) is considered optimal from the logistics point of view because it yields better results in transportation and warehousing efficiency. The strength of standardised packaging is that it makes it easier to develop efficient logistics systems because it places similar demands on transport and material handling equipment (Sonneveld 2000). However, standardisation may also lead to less adaptability with regard to change (Jahre and Hatteland 2004). Thus, when setting standard specifications for packaging, it is important to anticipate future changes of the packaging context and the permanence of these specifications (Koehorst et al. 1999). From a logistics point of view, it is indispensable to consider the storage and distribution of products, optimising the number of vehicles and routes, and reducing waste trips. In recent years, the traceability of packages, and thus of products, during distribution has been of fundamental importance. Packaging traceability technologies (e.g., Bar code, Radio Frequency Identification—RFID) allow for the identification of package positions in real time and continuously, increase of packaging information, reduction of shipment delays and tracking of lost shipments. Package identification also protects against theft and product manipulation during distribution. Theoretically the application of RFID technology to product packages could lead to a more detailed knowledge of the impact of real-time data, i.e., the identification of products in real time, the evaluation of the travelled time and distance travelled in a logistics system. At the end of the chapter, we will thoroughly discuss the use of RFID technology traceability in packaging logistics.

13 Packaging Logistics

281

Fig. 13.5 Foldable plastic fruits and vegetables crate (http://www.freshplaza.it/images/2012/120 6/CPR3.JPG)

Another relevant aspect to consider regarding packaging logistics is related to reverse logistics. Reverse logistics is the term used to describe the return flows of packaging and shipping materials from the retailer to the manufacturer. These flows are becoming relevant because of the impact on the environment (packaging at the end of use represents waste materials) and can be business for providers collecting packaging on the market and low-cost raw material for manufacturers. Additionally, the reverse logistics of packaging must be optimised in terms of transportation efficiency (Fig. 13.5). Since the 1990s, the environment has become an increasingly important consideration in the packaging system. Packages should be developed by using as little material as possible to reduce waste and minimise pollutant emissions when packaging waste is incinerated or landfilled. The reduction of waste and consequently of the environmental impact of packaging became possible by selecting recyclable materials and designing packaging that could be re-usable. In this way, the disposal of packaging could decrease the negative environmental effects produced by pollutant emissions and reduce waste volume. The re-use of packaging concerns its repeated use for a similar or for a different function (e.g., a cardboard package could become a box, a bag or a shelf). This aspect may be of fundamental importance in several sectors, such as the humanitarian logistics field, where it is important to use all transported materials in the field and the packages of humanitarian supplies, to save money and help as many people as possible (Regattieri et al. 2016). Another relevant environmental aspect to consider is sustainability. The environmental sustainability is the rate of renewable resource harvest, pollution creation, and non-renewable resource depletion that can be continued indefinitely. Environmental sustainability is an interesting field because it operates to protect the environment and preserve scarce resources for both present people and future generations, improving efficiency and optimising continuously the environmental performance of the packaging system.

282

A. Regattieri et al.

The fifth fundamental packaging key driver concerns the cost evaluation. The final target is the satisfaction of all requirements discussed before at minimum cost. The optimal setting of the entire “packaging system” (in terms of materials, shape, transportation, stocking, marketing feeling, environmental footprint) can create important savings and benefits for the companies that consider and analyse it. Packaging involves several industrial areas, and several packaging cost parameters should be considered. They are related to cost of engineering, purchasing, transportation, warehousing, cost related to reverse logistics, and cost of disposal. The total packaging cost estimation is an articulated issue that normally involves several parties (suppliers, product manufacturers, and public environmental agencies) (Regattieri and Santarelli 2013).

13.4 Supply Chain Packaging Costs Evaluation and Unit Load Design (Advanced) As underlined in the previous sections, a packaging system has numerous drivers and implications along the supply chain in marketing, production, logistics, purchasing and transportation. Evaluating the total costs of a packaging solution is necessary to define and improve the optimal packaging system management. The study conducted by Regattieri and Santarelli (2013) and supported by many authors in a literature analysis, reveals that most companies do not estimate total packaging costs, and shows the lack of a complete function for calculating the total cost of packaging in a company. Case Study: IKEA, the Tea Candle Case (Gustafsson et al. 2005)1 IKEA, a Swedish multinational group of companies and a worldwide leader in ready-to-assemble home furniture, appliances and home accessories, was founded in 1943. Packaging and logistics are important factors for the company’s success. Over the years, IKEA packaging products have distinguished them for their original flat package and home assembly concept. IKEA has continuously searched to improve efficiency in packaging and in the entire supply chain, though every single little improvement should be helpful for the business. For this purpose, in 2002, during a packaging efficiency project, it was found that GLIMMA, a bag containing tea candles, had more air than any other package. Moreover, GLIMMA tea candles were one of the most sold products, so an efficiency of packaging would result as a significant benefit for the company.

1 https://lup.lub.lu.se/search/ws/files/6186134/626088.pdf.

13 Packaging Logistics

283

The original first package held 100 candles in a simple soft plastic bag (see left figure below). The bags were packed in large cardboard containers (secondary package) placed on Euro-pallets (1200 × 800 mm), offering a display function. The soft plastic bag was difficult to handle and also to expose. The store floor space utilisation by the pallet and the display functions of GLIMMA were not as good as desired. In November 2002, a project was initiated to investigate the potential for improvement in the product, packaging and distribution. In February 2003, the development team identified a solution, which was expected to fulfil all the technical proprieties of the tea candles. Instead of selling tea-candles in a simple plastic bag, the IKEA team decided to pack candles in a tidy rectangular shape, in different levels. Groups of 100 candles were packed with shrink-thermo-wrap plastics. The solution needs created a new customized machinery for packaging candle, built by a German company already a leader in the production of machines in candle industry (Fig. 13.6). Before the study the cardboard on each pallet contained 252 GLIMMA packs, each with 100 tea candles. The new tidy system held 360 pack of 100 candles, more than 40% improvement of product load on each pallet. The study reduced the number of pallets needed for transporting the GLIMMA products in IKEA stores from 59,524 to 41,667 (approx. −30%). Consequently, this improvement reduced also the number of shipping candles from warehouse to stores. It resulted in lower logistic costs and less environmental impact. It actually produced 21% less CO2 emissions from fossil fuel used in the vehicle journeys each year. Clearly the new packaging solution, with more volume-efficiency, decreased the use of packaging materials: plastics for first package and cardboard boxes for secondary. These savings proved that it was possible to increase the profit margin, as the price for 100 tea candles remains the same as before. Various authors have developed mathematical models, considering many cost parameters regarding the packaging system (primary, secondary, and tertiary packages and accessories) along the entire supply chain of a manufacturing company (Regattieri and Santarelli 2013). In general, a Supply Chain packaging model represents added value for companies seeking to estimate the total costs of their packaging system and, consequently, its impact on total company costs. Moreover, it may be possible to determine overlooked and oversized packaging factors. The former should be introduced in the calculation of the total packaging costs, while the latter could be reduced or eliminated. The manufacturing company can rent or purchase packages (primary, secondary, and tertiary and accessories) and raw materials (if the manufacturer produces packages internally) from the supplier. When goods arrive, they are received in the manufacturer’s receiving area, sorted and stored in the warehouse. If the company must

284

A. Regattieri et al.

Fig. 13.6 First packaging solution for tea candles (http://www.route79.com/assets/images/29120 4-14.jpg); final packaging reduction after the research (http://www.canberrasupplies.co.uk/ekmps/ shops/canberra2/images/300-x-ikea-glimma-tealight-small-candles-tea-lights-31628-p.jpg)

produce the packaging, the raw materials are chosen and brought to the manufacturing area, where packages are made and subsequently stored in the warehouse. The raw materials not used during the manufacturing stage are returned to the warehouse, creating a reverse flow of materials. When the finished products are produced, the packages are chosen from the warehouse and brought to the manufacturing area. The packages not used during the manufacturing stage are also returned to the warehouse, creating again a reverse flow. The finished products are packed, put onto a pallet, and delivered to retailers. In the following pages, the model suggested by Regattieri and Santarelli (2013) is discussed. The model considers the possibility to re-use packages after the delivery of finished products to the final customers and the possible disposal of packages if they are damaged. Moreover, the model considers the possibility for the manufacturer to make a profit on sub-products derived from the disposal of packages and/or from the sale of tertiary packages to the final customers. Tables 13.3, 13.4 and 13.5 in the annex, at the end of the chapter, describe indices, variables and cost parameters used in the following model. Equation (13.1) introduces the general formula of the model. CTOT  CENG + CORD + CPUR + CRENT + CEXT TRAN + CREC + CCOND + CINT TRAN + CSTOCK + CPICK + CINT TRAN 1 + CMAN + CREV 1 + CINT TRAN 2 + CSTOCK 1 + CPICK 1 + CINT TRAN 3 + CREV 2 + CRE−USE + CDISP − RSUB − RUDC

(13.1)

Equation (13.2) presents the mathematical model, with each cost parameter in detail.

13 Packaging Logistics

CTOT 

 4 m 

285

 CENG it

 + NORD ·

i1 t1

 +

 +  +  +  +

4  m 

 CORD it

i1 t1

4  s  m 





CPUR nit · (xnit + ynit ) +

n1 i1 t1 4  s  m 



CEXT TRAN nit · NEXT TRAN nit 

CCOND it

 +

i1 t1 4  m 

+ 

4  m 

+ 





CSTOCK it · (xit + yit + wit ) +

+

CINT TRAN 1 it · NINT TRAN 1 it

+

+

4  m 

CREV INT TRAN 1 it · NREV INT TRAN 1 it 

CINT TRAN 2 it · NINT TRAN 2 it 

 +





CPICK 1 it · xit + yit + wit

 

CMAN it · xit





+

+

4  m 

 CREV INT COND1 it

i1 t1 4  m 

 

CSTOCK 1 it · xit

i1 t1 4  m 

 CINT TRAN 3 it · NINT TRAN 3 it

i1 t1

 

4  m 

· NREV INT TRAN 2 it

+

4  m 

 CREV INT COND2 it

i1 t1



CREV EXT TRAN rit · NREV EXT TRAN rit

r1 i1 t1 4  m 





CREV EXT COND it

i1 t1



CPICK it · xit

 

CREV INT TRAN 2 it i1 t1  q 4 m 

 +

 

4  m 



i1 t1

+

CREC it

i1 t1

i1 t1





i1 t1



i1 t1

4  m 

4  m 

CINT TRAN it · NINT TRAN it

i1 t1

4  m 

4  m 



+

i1 t1



CRENT nit · wnit

i1 t1

i1 t1 4  m 



n1 i1 t1

n1 i1 t1 4  m 

4  s  m 

4  m  i1 t1

+

4  m 

 CDISP it

i1 t1

 RSUB it · rit





 q 4 m 

RUDC rit · urit



(13.2)

r1 i1 t1

This model is a complete tool for companies to analyse the total packaging costs, to understand packaging cost reductions after package improvements, and consequently to minimize the impact of total packaging cost on total company cost.

286

A. Regattieri et al.

13.5 Unit Load Design The movement of materials and product along the supply chain is based on unit load equipment. The formation of the unit load is one of the most important elements of logistical systems. It affects different terms of the total Supply Chain cost model discussed in the previous section. Unit loads equipment normally used in the Supply Chain are: pallets, skids, tote pans, bins, baskets, racks, cartons, bags, bulk load containers, and crates. Moreover, there are several accessories to help stabilise a load, such as straps, tape, glue, shrinkwrap and stretch-wrap (Fig. 13.7). The pallet is the most used interface (packaging unit) in the logistics distribution system. A pallet is a raised platform for lean suppliers and facilitates loading, transportation and storage in units using forklift truck. Most common pallets are made from wood, but for particular operations are also constructed of metal, fibreboard, plastic and paper. To facilitate the spreading of pallets between actors of the supply chain and to encourage a standardization in Europe, two different standards of pallet measure respectively 800 × 1200 mm and 1000 × 1200 mm were adopted. EUR are four-way pallets, made for lifting from any of the four side. However, there are various types of pallets including two-way double wing, twoway flush, two-way reversible flush, two-way single wing, four-way non-reversible flush, block type, foam-padded plywood pallet, and single face pallet. The definition of the optimal configuration both for a full pallet (containing only one product) and a mixed pallet (containing different products and packages) is an important issue that can have a significant impact on the supply chain. Currently, many software programs able to optimise the pallet load, considering dimensions, weights, and physical compression, are available (Fig. 13.8). Several of these software programs assist practitioners in creating the entire packaging chain, starting from the optimum primary package size, continuing with the pallet configuration, and ending with the truck/container placement, while providing stacking strength recommendations for homogeneous products based on their dimensions and other shipping requirements. These suites permit “what if” scenarios, such as adding or removing an intermediate pack, changing carton dimensions, adding overhang to the pallet load or limiting the number of layers for each load.

Fig. 13.7 Example of pallet, tote pans, bins and bags (http://ulfe.wikia.com/wiki/Unit_Load_For mation_Equipment_Wiki#cite_note-11)

13 Packaging Logistics

287

Fig. 13.8 Pallet optimisation approach: mixed optimised pallets [Courtesy of Logen Solutions (http://www.logens olutions.com/)]

13.6 New Challenges in Packaging 4.0 (State of the Art) Packaging is a multidimensional function that plays a fundamental role in organisations looking to achieve successful management of a Supply Chain. In the future, companies will face big challenges in this area. For example, the packaging of the future will be made from materials that are not only renewable but also smart. Scientists and engineers are currently working with interactive materials for packages that change their appearance and form in response to various stimuli, have functional pressure and act as bio-based barriers for food packaging. Generally, packaging materials are assumed to have a relevant role in the current industrial context due to the need to reduce the quantity of materials while maintaining the same level of protection of products. There are several other areas in which companies are being faced with interesting challenges, including the high-speed development of e-commerce, the increase of the need of product traceability and the role of intelligent packaging.

13.7 E-commerce Over the past twenty years, another commercial revolution has changed the world, beginning with the rise of the Internet and the birth of e-commerce. E-commerce changed our life, our habits, customer and seller relations, and especially manufacturing systems, which have passed from mass production to demand-driven. Ecommerce is one of the most important developments that emerged from the presence of the Internet. Shopping online is common in today’s society. According to statistics, the number of China’s online shoppers reached 413 million, accounting for approximately 60% of the total number of Internet users, by the end of December 2015. Rapid growth of the online shopping scale has promoted the rapid development of green (electricitybased) business logistics, including self-built logistics and the third-party express delivery market (Wang and Hu 2016). Customisation has become necessary thanks to the new Just-In-Time manufacturing system, which within an organisation supports functional activities such as

288

A. Regattieri et al.

marketing, purchasing, design, production, sales and distribution, human resource management, warehousing and supplier development. Changing also affects packaging, which has gone from being an essential part of products to a mere protection and logistic tool. Packaging functions have had to develop along with the new requirements enforced by e-commerce (i.e., increased amounts of packaging materials for each product, an increased need to protect items, end-life management, environmental sustainability, etc.). The advent and increase of e-commerce has had a significant impact on packaging systems. There are three main issues to discuss regarding the packaging system for e-commerce: logistics, marketing and the environment. The primary role of packaging in e-commerce is to protect, preserve and contain goods (INCPEN 2012; Hyde 2013). Adequate protection of goods can be ensured by selecting proper packaging materials and accessories and packaging design, taking into account the possibility that goods do not meet customer expectations and need to be re-packed and returned (Korzeniowski and Jasiczak 2005). One of the primary jobs of any package is to protect the product; this is especially important for e-commerce shipping cases. These packages must be designed to be sufficiently durable to withstand the often complex automated and manual supply chains involved in delivering a product to a consumer’s doorstep (Mohan 2016). Adequate protection of goods can be ensured by selecting proper packaging materials and accessories and packaging design, taking into account the possibility that goods do not meet customer expectations and need to be re-packed and returned. Some important factors that should be considered when choosing packaging materials include the strength of the packaging, weight, volume, and value of the items, and whether the package will be subject to moisture or other adverse conditions. Wang and Hu (2016) proposed a classification of packaging used in the ecommerce market, i.e., express packaging. They describe express packaging as comprising three parts: outer packaging, internal fillers and express way-bill. Outer packaging plays a major role in protection, generally including corrugated cardboard boxes and waterproof plastic bags, that is chosen in relation to shipped items. Usually, the outside will be supplemented and reinforced by a large amount of tape sealing. Companies such as Amazon use external tape as a seal of packaging: the tape presents the company logo and ensures packaging closure. Internal fillers can relieve the damage of goods, reducing pressure, friction, vibration, impact and other external damage. Internal fillers include a pluri-ball sheet, bubble bag, plastic foam, paper fill products, and dunnage bags. Flexible packaging products such as these will continue to dominate due to their cost efficiency and their ability to package a wide variety of goods. Finally, express way-bill, which is a contract of carriage and receipt of credentials, can indicate the information of acceptance and goods. Combined with the application of the existing bar code technology, transportation and distribution has become accurate and efficient. The right packaging material can make the difference between successfully shipping a fragile item or having it arrive in pieces (Sullivan 2016). PMMI, the US Association for Packaging and Processing technologies, reported that the consequences of a damaged product can be measured in both dollars and in brand reputation: “The cost of replacing a destroyed item can

13 Packaging Logistics

289

be up to 17 times the cost of shipping, and negative website reviews resulting from the destruction of an item can take months to counterbalance with positive ones” (Wang and Hu 2016). Other functions, e.g., ease of processing and handling, as well as transport, storage, convenience and re-use, are all affected by the packaging system. Moreover, several electronic tools, such as Electronic Data Interchange (EDI, i.e., the structured transmission of data between organisations by electronic means), can facilitate the management of online packaging from a logistics point of view. EDI enables minimal stocks to be held with consequent savings in storage, insurance, warehousing and labour costs (Gunasekaran et al. 2002). One of the biggest challenges of an e-commerce business is the so-called last mile, i.e., home delivery service for the customer. Frequent and numerous home deliveries may create incentives for manufacturers to take back their packaging, to use refillable or recyclable packaging materials, or to study new and cheaper home deliveries solutions (e.g., drone deliveries) to become increasingly cost-effective. The marketing of products has changed due to the Internet and e-commerce. The function of packaging a product in an attractive manner becomes less important. The more customers shop online, the less important shelf presentation will become. In a marketplace or in online shop, a buyer cannot touch the product nor directly see the package, while other packaging characteristics, such as protection and reusability for an efficient return of products, have more importance. The changing role of packaging for e-commerce regarding the purchase of a product makes it desirable and possible to pay more attention to the consumer perception of a brand while the user is using it and less attention to its shelf presentation. According to the main purposes fulfilled by packaging (Regattieri et al. 2014b; Wang and Hu 2016), the main packaging requirements that a company should consider before starting an e-commerce business are as follows: Protection: goods protection is the most relevant packaging function because products must arrive in good condition for consumers. Products have to be protected from mechanical, chemical and biological damage (Korzeniowski and Jasiczak 2005), using also some packaging accessories such as ball up paper, pluri-ball plastics, air pillows, polystyrene chips, and others; Handleability: the ergonomic aspect, i.e., everything related to adaptation to the human physique and behaviour when using the product, must be considered. To confirm this, the empirical study conducted regarding the packaging perception of end consumers stated that the main requirement that packaging should guarantee is the handleability (e.g., easy handling, easy opening, user-friendly, etc.); Security: packages must ensure secure shipping. It could be necessary to install identification technologies, such as Radio Frequency Identification (RFID) tags or barcodes, in packages to reduce thefts, increase security, and minimise time spent on the traceability of products; Respect for the environment: e-commerce produces more materials waste than traditional commerce because usually users buy more orders in smaller quantities, and because in addition to traditional first product package, shipment needs one or more layer of package for transport protection. To have the minimum environmen-

290

A. Regattieri et al.

tal impact, it could be necessary for companies to recycle packages and minimise dangerous substances in emission when packaging waste is disposed; Re-use: more and more companies have started to re-use packages to ship products to end consumers, minimising both the environmental impact and costs. The re-use of packages could also increase customer satisfaction thanks to the low environmental pollution produced. Another e-commerce packaging consideration is the ability to attract consumers’ attention and piquing their curiosity about products, which are important factors for analysis to increase the potential development of packages for online shopping. Further research should focus on the definition of a packaging solution for different sectors (e.g., luxury products, high-tech products, etc.). Thus, it might be interesting to compare packaging solutions developed for online markets of low- and high-value goods.

13.8 Packaging and Traceability Traceability is similar to the history of a product in terms of the direct properties of that product and/or properties that are associated with that product once these products have been subject to particular value-adding processes using associated production means and in associated environmental conditions. The information concerning relationships at the origin may be used upstream in the supply chain (e.g., in the ordering process to define the requirements of an ordered product) or downstream (e.g., in delivery processes to specify the characteristics of products). Additionally, the information can be used for reporting purposes, either in the supply chain or for third parties (Regattieri et al. 2007). Traceability is a concept related to all products and all types of supply chains. Currently, in an economic system in which companies compete against each other in an environment largely founded on customer satisfaction, traceability is an indispensable instrument in obtaining market consensus. Direct benefits are supply chain optimisation, product safety, and market advantages (marketing advantages/competitive business advantages). An efficient and effective system transmitting accurate, timely, complete, and consistent information regarding products through the supply chain can significantly reduce operating costs and increase productivity. At the same time, such a system contains many product safety elements: it makes consumers safer by providing detailed information regarding where an item comes from, what its components and origin are, and its processing history (Regattieri et al. 2007). A product traceability system requires the identification of all the physical entities (and locations) from which the product originates, i.e., where it is processed, packed, and stocked; this includes every agent in the supply chain. Current technical and operative resources available are fundamentally the alphanumerical code, bar code, and radio-frequency identification; these resources are placed on or drowned inside (through innovative RFID systems) the product package (primary, secondary or tertiary).

13 Packaging Logistics

291

The use of alphanumerical codes and barcodes for internal and external supply chains has been well known for decades. In recent years, the application of RFID has attracted considerable interest among scientists and managers. Many manufacturers (e.g., Tesco, Marks & Spencer and other retailers) have already started to use RFID successfully (Regattieri et al. 2014a), installing the tags into secondary packages to minimise costs and increase the speed of the products identification phase (Singh et al. 2009). In recent years, Wal-Mart has invited its eight main suppliers, i.e., Gillette, Hewlett-Packard, Johnson & Johnson, Kimberly-Clark, Kraft Foods, Nestlé Purina PetCare Co., Procter & Gamble and Unilever, to pilot case and pallet-level RFID during their procurement processes to increase transaction accuracies and decrease inventory errors (Cui et al. 2017). An RFID system consists of three basic components: a tag, i.e., transponder, a reader/writer, and the host computer or processor. The readers communicate with tags and collect information regarding the positions of the objects hosting the tags. Tags are usually placed on the product package (primary and/or secondary and/or tertiary) or involve different innovative solutions, e.g., being drowned inside the package material. Some of the advantages of the RFID system are as follows: it does not require line-of-sight to capture data, thus saving time and labour by eliminating the need to manage products and identify them (Hellström and Wiberg 2009; McCarthy et al. 2009); it is able to read the contents of an entire unit load in seconds, reducing time and labour (Hellström and Wiberg 2009); it is able to track more than one position simultaneously (Gezici et al. 2005). The application of RFID to the packaging system allows more frequent and automated identification of packages, increasing the accuracy of the system and customer service level, reducing the labour and time needed to identify packages, eliminating manual pallet scanning and enabling near real-time visibility, which in turn facilitates the coordination of activities within and between processes (Clarke et al. 2006; Singh et al. 2009). On the other hand, the RFID method is an expensive solution, though this limitation could be overcome by better performing RFID systems (Hellström and Wiberg 2009). Although the major focus has been on the cost of tags, in relation to their size and types, another possible limitation of the RFID system is the interference caused by some equipment in factories (McCarthy et al. 2009). Metal and water have a detrimental effect on radio waves: metal reflects the radio waves and water absorbs and/or reflects them. This limitation could be overcome by carefully chosen tag orientation (Seidel and Donner 2010). Several authors have described the main industrial applications of RFID systems for companies, focusing on logistics (e.g., real-time identification of the position of materials and packages), healthcare, security and identification systems. Accurate material flow identification can increase the knowledge and the measurement of the entire supply chain (Allesina et al. 2010). Many companies need to track packages, first without the product and then with products inside, to know the real path (and cost) of their material flows, allowing control of the work in progress, and to reduce costs of the system.

292

A. Regattieri et al.

Airplane Maintenance: Tracking of Spare Parts in Air France KLM2 KLM Engineering & Maintenance is the maintenance department of Air France KLM airplanes fleet. Every day KLM E&M manages a wide logistics network for tracking airplane spare parts shipped all over the world. From 300 facilities, engineers maintain, repair and overhaul about 1,500 airplanes for 150 different airline companies. Many parts are removed, repaired or replaced for planned maintenance. These parts are shipped in special tracked packages from Nefab, a Swedenbased multinational packaging company and leading provider of packaging solutions. By tracking their shipping, KLM can monitor parts inside, as well as track packaging that gets lost. Tracking packaging allows to improving spare parts management and also reducing aircrafts maintenance down-time. Tracking spare parts in aircraft maintenance is essential, especially to ensure the connection between the correct spare parts with the types and models of aircrafts. This innovation is important in strengthening flight safety. KLM worked with Nefab, to find a solution and together developed the Aviation Packaging Information System (APIS) a web-based RFID system for tracking flows (inbound and outbound) of spare parts between Nefab and KLM maintenance department. To track packages, fixed RFID readers have been installed in warehouse and shipping truck. Where it was impossible to install a fixed reader, a mobile RFID handhelds is used instead. Furthermore, in Nefab facilities several smartspots have been installed, a sort of high performance RFID antenna that can capture data regarding packages that Nefab ships to KLM. Additionally, several different smartspots have been installed in KLM’s engineering department, logistics centre, warehouse and shop, where maintenance is performed. All packaging used by KLM for the spare parts was tagged with one or more RFID tags. The warehouse management and tracking software updates automatically if a smartspot captures the tag ID of a package. Results: Due to this RFID solution, KLM could track and trace the location of equipment. The packaging costs were reduced by 50%. Additionally, the system is virtually 100% hands-free and can be paperless. This means KLM E&M’s employees no longer need to manually record when packages arrive, which results for KLM in faster processes and additional savings.

2 http://www.cisper.nl/case-studies/rfid-case-study-klm-rfid-in-packaging/.

13 Packaging Logistics

293

Packaging incorporating RFID technology is usually referred to as smart packaging (also called active or intelligent packaging) and is commonly used to describe packaging with different types of value-adding technologies, e.g., placing a smart label or tag inside a package. The term smart packaging was used by Yam in 2000 (Yam 2000) to emphasise the role of packaging as an intelligent messenger or an information link. Smart packaging is another interesting area with a significant potential to gain efficiencies and decrease system-wide costs for the packaging industry in the near future as described in the next section.

13.9 Smart (Intelligent) Packaging According to the Smart Packaging Journal (2002), smart packaging is described as packaging that employs features of high added value that enhances the functionality of the product; its core is responsive features. Smart packaging is often used to refer to electronic responsive features where data is electronically sensed on the package from a distance, using an automatic identification system such as the RFID technology. However, changes in consumer preference and the increasing desire for safeguarding the quality, health and safety of food and consumable products have led to the development of intelligent packaging technologies. Active packaging refers to the incorporation of additives into a package, with the aim of maintaining or extending the product quality and shelf life. The intelligent systems are those that monitor the condition of packaged food to give information regarding the quality of the packaged food during transportation and storage (Biji et al. 2015). In the traditional point of view, packaging is a barrier and protection between products and the external environment along the entire supply chain. A surge in temperature, heat, cold, moisture, sun light, pressure, gaseous emission and microorganisms could deteriorate the quality of products. Spoilage of food is a change in a food product that makes it unacceptable for consumption from a sensory point of view. Microbial spoilage is the most common cause of food loss, affecting 25% of all foods produced globally (Lorite et al. 2017). Some engineered packaging films contain enzymes, anti-bacterial agents, scavengers, and other active components to help control food degradation and extend shelf life. To increase product quality and to reduce food waste and meet the needs of modern consumer demand regarding the quality of food products, it is necessary to monitor the conditions of food during shipment or use intelligent packaging, such as the active part of it, to maintain or extend the product quality and shelf life. However, intelligent packaging development is not easy; it is highly complex because it requires multi-sector knowledge and multi-disciplinary work in different fields, such as food science, material science, and chemical and electric engineering.

294

A. Regattieri et al.

Fig. 13.9 Temperature sensor on fresh fish packaging (http://iot-spain.com/?p=3504&lang=en). Embedded microchip on pharmaceutical packaging application to remind people to take medication and record response to treatment (http://www.hightechfinland.com/direct.aspx?area=htf&prm1=7 16&prm2=article)

Various strategies are used for the preservation of food in packages and for prolonging shelf life, e.g., temperature control, oxygen removal (or oxygen scavengers), antimicrobial substances, moisture control, flavour or odour, ethylene scavengers, absorbers and releasers, and the addition of chemicals such as salt, sugar, carbon dioxide or natural acids or a combination of these with effective packaging. Current developments in tags have made tags with humidity, temperature, pressure and moisture monitoring applications able to understand information to manage products and supply chains in effective ways (Fig. 13.9). Active and intelligent packaging represent exciting opportunities not only for product safety and quality but also for Supply Chain effectiveness. Some experts (Cushen and Cummins 2017) believe that the next round of technology in packaging will include nanotechnologies that will allow new compounds such as novel antimicrobials and gas scavengers to be included in packaging films. The advancement of electronic devices will also help drive the innovation of active and intelligent packaging. The use of active and intelligent packaging will likely become more popular as more technologies make their way to the market, and innovative packaging in active and intelligent systems will become more common place. Perhaps active and intelligent packaging will completely replace traditional packaging itself. As Paul Gander of Food Manufacture Magazine states, “[..] The trend is towards less packaging, and what there is will be more interactive. Whether 2020 will see packs which literally walk off the shelf is quite another matter” (Gander 2007; Huff 2008; Maisanaba et al. 2016). Another research path for intelligent packages addresses the automatic changing of packaging shape in reaction to different stimuli (e.g., heat, mechanical stress, electrical stress, etc.). In the near future, we can have a package that opens itself when the temperature (in the oven) is just right and the food is ready to be served by studying mechano-active materials that will change shape when exposed to high temperature.

13 Packaging Logistics

295

13.10 Further Reading On packaging logistics (the basic section of chapter) several interesting papers have been written in the last years. An interesting reading on the main role of packaging within the supply chain was published by Chan et al. (2006). Twede (2005) analyzed the use of wooden barrels, one of the most popular packaging through the Centuries. Barrels not only changed the course of logistics, but also history, definitively opening the route to worldwide deliveries. The design of environmentally sustainable logistics has emerged as one of the most influential challenges in recent years. An in-depth discussion is provided by Abbasi and Nilsson (2016). On the same topic, Swedish researchers have recently started to study and to propose various interesting solutions. From a packaging point of view, researchers are continuously looking for improvements, new materials and new technologies for innovative packaging solutions, mainly focused on the environment and on packaging biodegradability (e.g. edible water “bottle”, etc.). A good contribution on this field, in particular on biodegradable shear oriented packaging, is discussed by Frackowiak et al. (2016). Further developments include advances in RFID technologies (Zou et al. 2014; Huang et al. 2016). In few years’ time, these technologies probably will be offered as commercial solutions for daily packaging traceability. Advances in nanotechnologies are also expected to impact future packaging logistics. Mlalila et al. (2016) discuss new developments in particular in the food sector.

Annex Tables 13.3, 13.4 and 13.5.

Table 13.3 Indices of the model Index Domain i

Description

1, …, 4

Level of package: i  1 (primary package) i  2 (secondary package) i  3 (tertiary package) i  4 (accessories)

t

1, …, m

Different packages for each level i

n

1, …, s

Suppliers

r

1, …, q

Retailers

296

A. Regattieri et al.

Table 13.4 Variables of the model Variable Units

Description

Domain

xnit

(pieces/year)

Quantity of raw materials bought by the company from the supplier n to produce package i of type t

i  1, …, 4; t  1, …, m; n  1, …, s

x it

(pieces/year)

Quantity of package i of type t produced by the manufacturer company from raw materials

i  1, …, 4; t  1, …, m

ynit

(pieces/year)

Quantity of package i of type t bought by the company from supplier n

i  1, …, 4; t  1, …, m; n  1, …, s

wnit

(pieces/year)

Quantity of package i of type t rented by the company from the supplier n

i  1, …, 4; t  1, …, m; n  1, …, s

rit

(pieces/year)

Quantity of disposed package i of i  1, …, 4; type t from which the company has t  1, …, m a profit from sub-products

urit

(pieces/year)

Quantity of package i of type t sold i  1, …, 4; by the company to the retailer r t  1, …, m; r  1, …, q

NORD

(orders/year)

NEXT TRAN nit

(trips/year)

Number of orders for buying raw materials and/or packages i of type t Number of trips of raw materials i  1, …, 4; and/or packages i of type t from t  1, …, m; the supplier n to the manufacturer n  1, …, s

NINT TRAN it

(trips/year)

NINT TRAN it

(trips/year)

NINT TRAN it

(trips/year)

NINT TRAN it

(trips/year)

Number of trips of raw materials and/or packages i of type t from the manufacturer’s receiving area to the warehouse Number of trips of raw materials i of type t from the warehouse to the manufacturing area to produce packages from xit

i  1, …, 4; t  1, …, m

Number of trips of packages i of type t produced by the manufacturer and transported from the production area to the warehouse Number of trips of packages (produced/bought/rented) i of type t from the warehouse to the production area to support finished products

i  1, …, 4; t  1, …, m

i  1, …, 4; t  1, …, m

i  1, …, 4; t  1, …, m

(continued)

13 Packaging Logistics Table 13.4 (continued) Variable Units NREV INT TRAN it

(trips/year)

NREV INT TRAN it

(trips/year)

NREV EXT TRAN rit

(trips/year)

297

Description

Domain

Number of trips of packages i of type t not used during the production of finished products and transported from the manufacturing area to the warehouse Number of trips of the quantity of raw materials i of type t not used during the production of packages and transported from the manufacturing area to the warehouse Number of trips of packages i of type t from the retailer r to the manufacturer

i  1, …, 4; t  1, …, m

Table 13.5 Cost parameters of the model Parameter Nomenclatures Units (e/year)

i  1, …, 4; t  1, …, m

i  1, …, 4; t  1, …, m; r  1, …, q

Description

CENG

Cost of Engineering

CORD

Cost of Purchase (e/order) Order

Cost of managing the internal purchase orders if the manufacturer produces the packaging internally; otherwise, it represents the purchase orders for buying and/or renting packaging from suppliers. It includes the labour costs for making the order

CPUR

Cost of Purchasing

(e/piece)

Purchase cost of raw materials (to produce packaging) and/or packages

CRENT

Cost of Rent

(e/piece)

Cost to rent packages

CEXT TRAN

Cost of External (e/travel) Transport

CREC

Cost of Receiving

(e/year)

Cost of studying each type of packaging and for making prototypes. It includes the labour costs of engineering the product

Cost of transporting raw materials and/or packages from the supplier to the manufacturer: it comprises labour costs, depreciation of vehicles (e.g., truck), and cost of the distance travelled Cost of receiving raw materials and/or packages. It includes the labour costs and depreciation of vehicles (e.g., truck, forklift) used to unload products (continued)

298

A. Regattieri et al.

Table 13.5 (continued) Parameter Nomenclatures

Units

Description

CCOND

Cost of Conditioning

(e/year)

Cost of sorting raw materials and/or packages before storing them in the warehouse. It includes the labour costs and depreciation of mechanical devices (if used), e.g., for unpacking and re-packing products

CINT TRAN

Cost of Internal Transport

(e/travel)

CSTOCK

Cost of Stocking (e/piece)

Cost of transporting raw materials and/or packages from the manufacturer’s receiving area to the warehouse. It includes the labour costs, depreciation of vehicles (e.g., forklift), and cost of the distance travelled Cost of storing raw materials and/or packages in the warehouse. It includes the labour costs and the cost of the space used for storing the packages

CPICK

Cost of Picking

(e/piece)

Cost of picking up raw materials from the warehouse for producing the packages. It includes the labour costs and depreciation of vehicles (e.g., forklift) for picking up the products

CINT TRAN

Cost of Internal Transport

(e/travel)

CMAN

Cost of Packages Manufacturing

(e/piece)

Cost of transporting raw materials from the warehouse to the manufacturing area to produce the packages. It includes the labour costs, depreciation of vehicles (e.g., forklift), and cost of the distance travelled Cost of producing packages internally; it includes the labour costs, depreciation of production plants and utilities (e.g., electricity, water, gas, etc.)

CREV

Cost of Internal Reverse Logistics

(e/travel)

Cost of transport for bringing the raw materials not used during manufacturing back to the warehouse. It includes: C1REV INT TRAN : the cost of transport for coming back to the warehouse. It comprises labour costs, depreciation of vehicles used (e.g., forklift), and cost of the distance travelled; C1REV INT COND : the cost of conditioning packages to make them re-usable. It comprises the labour costs and depreciation of mechanical devices (if used), e.g., for unpacking and re-packing products (continued)

13 Packaging Logistics Table 13.5 (continued) Parameter Nomenclatures

299

Units

Description

(e/travel)

Cost of transporting the packages produced by the company from the production area to the warehouse. It includes the labour costs, depreciation of vehicles (e.g., forklift), and cost of the distance travelled Cost of stocking packages produced internally by the company. It includes the labour costs and cost of the space for storing the packages

CINT TRAN

Cost of Internal Transport

CSTOCK

Cost of Stocking (e/piece)

CPICK

Cost of Picking

(e/piece)

Cost of picking up packages (produced/bought/rented) from the warehouse. It includes the labour costs and depreciation of vehicles (e.g., forklift) for picking up the packages

CINT TRAN

Cost of Internal Transport

(e/travel)

CREV

Cost of Internal Reverse Logistics

(e/travel)

Cost of transporting packages from the warehouse to the manufacturing area. It includes the labour costs, depreciation of vehicles (e.g., forklift), and cost of the distance travelled Cost of transport for bringing packages not used during the manufacturing of finished products back to the warehouse. It includes: C2REV INT TRAN : the cost of transport back to the warehouse. It comprises the labour costs, depreciation of vehicles used, and cost of the distance travelled; C2REV INT COND : the cost of conditioning packages to make them re-usable. It comprises the labour costs and depreciation of mechanical devices (if used), e.g., for unpacking and re-packing products

CRE-USE

Cost of Re-Use

(e/year)

Cost of re-using packaging after the delivery of finished products to the customer. It includes: CREV EXT TRAN : the cost of transport back to the company. It comprises the labour costs, depreciation of vehicles used (e.g., truck), and cost of the distance travelled; CREV EXT COND : the cost of conditioning packages to make them re-usable. It comprises the labour costs and depreciation of mechanical devices (if used), e.g., for unpacking and re-packing products (continued)

300

A. Regattieri et al.

Table 13.5 (continued) Parameter Nomenclatures

Units

Description

CDISP

Cost of Disposal (e/piece)

Cost of disposing of damaged packages during the manufacturing stage. It comprises the cost of disposal and the cost of transporting damaged packages from the company to the landfill (labour costs, depreciation of vehicles used (e.g., truck), and cost of the distance travelled)

RSUB

Gain from Sub-Product

(e/piece)

The parameter identifies the possible gain obtained from the disposal of damaged products

RUDC

Gain from Direct Sale of Pallet

(e/piece)

This parameter identifies the possible gain obtained from the sale of tertiary packaging to the final customer

References Abbasi M, Nilsson F (2016) Developing environmentally sustainable logistics. Exploring themes and challenges from a logistics service providers’ perspective. Transp Res Part D Transp Environ 46:273–283. https://doi.org/10.1016/j.trd.2016.04.004 Allesina S, Azzi A, Battini D, Regattieri A (2010) Performance measurement in supply chains: new network analysis and entropic indexes. Int J Prod Res 48:2297–2321. https://doi.org/10.1080/00 207540802647327 Beckeman A, Lu M, Lu C (2007) Past and present roles of packages in logistics and towards the future. In: Logistics research network 2007 conference proceedings Biji KB, Ravishankar CN, Mohan CO, Srinivasa Gopal TK (2015) Smart packaging systems for food applications: a review. J Food Sci Technol 52:6125–6135. https://doi.org/10.1007/s13197015-1766-7 Chan FTS, Chan HK, Choy KL (2006) A systematic approach to manufacturing packaging logistics. Int J Adv Manuf Technol 29:1088–1101. https://doi.org/10.1007/s00170-005-2609-x Chapman RL, Soosay C, Kandampully J (2003) Innovation in logistic services and the new business model. Int J Phys Distrib Logist Manag 33:630–650. https://doi.org/10.1108/096000303104992 95 Clarke RH, Twede D, Tazelaar JR, Boyer KK (2006) Radio frequency identification (RFID) performance: the effect of tag orientation and package contents. Packag Technol Sci 19:45–54. https:// doi.org/10.1002/pts.714 Cui L, Deng J, Liu F et al (2017) Investigation of RFID investment in a single retailer two-supplier supply chain with random demand to decrease inventory inaccuracy. J Clean Prod 142:2028–2044. https://doi.org/10.1016/j.jclepro.2016.11.081 Cushen M, Cummins E (2017) Emerging technologies in meat processing. In: Cummins EJ, Lyng JG (eds). Wiley, Chichester, UK, pp 265–284 Davis G, Song JH (2006) Biodegradable packaging based on raw materials from crops and their impact on waste management. Ind Crops Prod 23:147–161. https://doi.org/10.1016/j.indcrop.20 05.05.004 Dominic C (2005) Integrating packaging suppliers into the supply/demand chain. Packag Technol Sci 18:151–160. https://doi.org/10.1002/pts.684 Domnica D (2010) Review concerning the functions of packaging. 1:44–49

13 Packaging Logistics

301

Frackowiak S, Ludwiczak J, Leluk K, Kozlowski M (2016) New class of shear oriented, biodegradable packaging material. Compos Part B Eng 92:1–8. https://doi.org/10.1016/j.compositesb.201 6.02.036 Gander P (2007) The smart money is on intelligent design. Food Manuf xvxvi:2007 Gezici S, Tian Z, Giannakis GB et al (2005) Localization via ultra-wideband radios: a look at positioning aspects of future sensor networks. IEEE Signal Process Mag 22:70–84. https://doi.o rg/10.1109/MSP.2005.1458289 Gray V, Guthrie J. Environmentally packaging, pp 31–36 Gunasekaran A, Marri HB, McGaughey RE, Nebhwani MD (2002) E-commerce and its impact on operations management. Int J Prod Econ 75:185–197. https://doi.org/10.1016/S0925-5273(01)0 0191-8 Gustafsson K, Jonson G, Smith D, Sparks L (2005) Packaging logistics and retailers’ profitability: an IKEA case study Hellström D, Saghir M (2007) Packaging and logistics interactions in retail supply chains. Packag Technol Sci 20:197–216. https://doi.org/10.1002/pts.754 Hellström D, Wiberg M (2009) Exploring an open-loop RFID implementation in the automotive industry. In: ETFA 2009—2009 IEEE conference on emerging technologies and factory automation, pp 1–4. https://doi.org/10.1109/etfa.2009.5347254 Hermansson A (1999) Openability of retail packages. Packag Technol Sci 12:219–223. https://doi. org/10.1002/(sici)1099-1522(199909/10)12:53.0.co;2-d Huang Y, Xu Y, Qi S et al (2016) Recent patents on RFID-based logistics management systems, pp 26–36 Huff K (2008) Active and intelligent packaging: innovations for the future. Dep Food Sci Technol 1–13 Hyde J (2013) How to pack and protect an e-commerce shipment? http://www.practicalecommerc e.com/articles/3349-How-to-Pack-and-Protect-an-Ecommerce-Shipment INCPEN (2012) Packaging and the intern a guide to packaging goods for multi-channel delivery systems Jahre M, Hatteland CJ (2004) Packages and physical distribution: Implications for integration and standardisation. Int J Phys Distrib Logist Manag 34:123–139. https://doi.org/10.1108/09600030 410526923 Jönson G, Johnsson M (2006) Packaging technology for the logistician, 3rd edn, Department of Design Sciences, Division of Packaging Logistics, Lund University, Lund Klevas J (2006) Design for packaging logistics. Int Des Conf 269–276 Klevås J (2005) Organization of packaging resources at a product-developing company. Int J Phys Distrib Logist Manag 35:116–131. https://doi.org/10.1108/09600030510590309 Koehorst H, de Vries H, Wubben E (1999) Standardisation of crates: lessons from the Versfust (Freshcrate) project. Supply Chain Manag An Int J 4:95–101. https://doi.org/10.1108/13598549 910264770 Korzeniowski A, Jasiczak J (2005) LogForum. 1:1–7 Lockamy A (1995) A conceptual framework for assessing strategic packaging decisions. Int J Logist Manag 6:51–60 Lorite GS, Selkala T, Sipola T et al (2017) Novel, smart and RFID assisted critical temperature indicator for supply chain monitoring. J Food Eng 193:20–28. https://doi.org/10.1016/j.jfooden g.2016.06.016 Maisanaba S, Llana-Ruiz-Cabello M, Gutiérrez-Praena D et al (2016) New advances in active packaging incorporated with essential oils or their main components for food preservation. Food Rev Int 9129:1–69. https://doi.org/10.1080/87559129.2016.1175010 Matzler K, Hinterhuber HH, Bailom F, Sauerwein E (1996) How to delight your customers. J Prod Brand Manag 5:6–18. https://doi.org/10.1108/10610429610119469 McCarthy U, Ayalew G, Butler F et al (2009) The effects of item composition, tag inlay design, reader antenna polarization, power and transponder orientation on the dynamic coupling efficiency of

302

A. Regattieri et al.

backscatter ultra-high frequency radio frequency identification. Packag Technol Sci 22:241–248. https://doi.org/10.1002/pts.849 Mlalila N, Kadam DM, Swai H, Hilonga A (2016) Transformation of food packaging from passive to innovative via nanotechnology: concepts and critiques. J Food Sci Technol 53:3395–3407. https://doi.org/10.1007/s13197-016-2325-6 Mohan AM (2016) How e-commerce is changing the packaging landscape Olander-Roese M, Nilsson F (2009) Competitive advantage through packaging design. Propositions for supply chain effectiveness and efficiency. In: Proceedings of the ICED 09, 17th international conference on engineering design, pp 279–290 Olsson A, Györei M (2002) Packaging throughout the value chain in the customer perspective marketing mix. Packag Technol Sci 15:231–239. https://doi.org/10.1002/pts.585 Olsson A, Larsson AC (2009) Value creation in PSS design through product and packaging innovation processes. In: Sakao, Lindahl (eds) Introduction to product/service-system design. Springer, pp 93–108 Olsson A, Petterson M, Jonson G (2004) Packaging demands in the food service industry. Food Serv Technol 4:97–105. https://doi.org/10.1111/j.1471-5740.2004.00095.x Paine FA (1990) packaging design and performance, Pira inter Prendergast G, Leyland P (2009) Packaging, marketing, logistics and the environment: are there trade-offs? Int J Phys Distrib Logist Manag 26:60–72. https://doi.org/10.1108/02656710210415 703 Regattieri A, Gamberi M, Manzini R (2007) Traceability of food products: general framework and experimental evidence. J Food Eng 81:347–356. https://doi.org/10.1016/j.jfoodeng.2006.10.032 Regattieri A, Piana F, Bortolini M, Gamberi M, Ferrari E (2016) Innovative portable solar cooker using the packaging waste of humanitarian supplies. Renew Sustain Energy Rev 57:319–326. https://doi.org/10.1016/j.rser.2015.12.199 Regattieri A, Santarelli G (2013) The important role of packaging in operations management. In: Schiraldi M (ed) Operations management. InTech. https://doi.org/10.5772/540 73. https://www.intechopen.com/books/operations-management/the-important-role-of-packagin g-in-operations-management Regattieri A, Santarelli G, Gamberi M, Gamberini R (2014a) The use of radio frequency identification technology in packaging systems : experimental research on traceability. Packag Technol Sci 591–608. https://doi.org/10.1002/pts Regattieri A, Santarelli G, Gamberi M, Mora C (2014b) A new paradigm for packaging design in web-based commerce. Int J Eng Bus Manag 6:1–11. https://doi.org/10.5772/58825 Saghir M (2002) Packaging logistics evaluation in the Swedish retail supply chain. Lund University Saghir M, Jönson G (2001) Packaging handling evaluation methods in the grocery retail industry. Packag Technol Sci 14:21–29. https://doi.org/10.1002/pts.523 Saghir ML (2004) A platform for packaging logistics development: a systems approach. Department of Design Sciences, Faculty of Engineering, Lund University Sara R (1990) Packaging as a retail marketing tool. Int J Phys Distrib Logist Manag 20:29–30. https://doi.org/10.1108/EUM0000000000372 Seidel T, Donner R (2010) Potentials and limitations of RFID applications in packaging industry: a case study. Int J Manuf Technol Manag 21:225. https://doi.org/10.1504/IJMTM.2010.035433 Silayoi P, Speece M, Silayoi P, Speece M (2004) Packaging and purchase decisions. Br Food J 106:607–628. https://doi.org/10.1108/00070700410553602 Silva MA, Jourdes M, Darriet P, Teissedre PL (2012) Scalping of light volatile sulfur compounds by wine closures. J Agric Food Chem 60:10952–10956. https://doi.org/10.1021/jf303120s Singh J, Olsen E, Vorst K, Tripp K (2009) RFID Tag readability issues with palletized loads of consumer goods. Packag Technol Sci 22:431–441. https://doi.org/10.1002/pts.864 Sonneveld K (2000) The role of life cycle assessment as a decision support tool for packaging. Packag Technol Sci 13:55–61. https://doi.org/10.1002/1099-1522(200003/04)13:23.0.co;2-g Sullivan D (2016) Different types of packaging materials, pp 1–2

13 Packaging Logistics

303

Titus E, Ahearn GA (1992) Editorial review. Packag Technol Sci 4:243–250 Twede D (2005) The cask age: the technology and history of wooden barrels. Packag Technol Sci 18:253–264. https://doi.org/10.1002/pts.696 Verghese K, Lewis H (2007) Environmental innovation in industrial packaging: a supply chain approach. Int J Prod Res 45:4381–4401. https://doi.org/10.1080/00207540701450211 Wang F, Hu Y (2016) Research on green express packaging design under the electronic commerce. Open J Bus Manag 4:621–628. https://doi.org/10.4236/ojbm.2016.44063 Yam KL (2000) Intelligent packaging for the future smart kitchen. Packag Technol Sci 13:83–85. https://doi.org/10.1002/1099-1522(200003/04)13:23.0.co;2-7 Zou Z, Chen Q, Uysal I, Zheng L (2014) Radio frequency identification enabled wireless sensing for intelligent food logistics. Philos Trans R Soc A Math Phys Eng Sci 372:20130313–20130313. https://doi.org/10.1098/rsta.2013.0313 (2002) Why the use of RFID in smart packaging?

Chapter 14

Outbound Logistics and Distribution Management Matthias Klumpp and Sunderesh Heragu

Abstract Distribution and outbound processes are important for many companies because they directly connect them with the customers in a value chain. Market and customer demands relative to quality, speed as well as information and service orientation of logistics processes matter in terms of overall evaluation and satisfaction. At the same time, there are significant cost advantages or disadvantages. In the trade and retail sector, this operational field is especially of high strategic importance and closely connected to e-commerce or multi-channel strategies. It can be said that all the excellence as well as product and service quality built up throughout the value chain can be delivered or destroyed within these last miles of distribution, point of sale, and customer contact. This chapter outlines the core definitions and objectives for outbound logistics and distribution management (Sect. 14.1) before providing an extensive case study for this specific topic (Sect. 14.2). It then provides the operational concepts for distribution in Sect. 14.3 (basic level). In Sect. 14.4 (advanced level), it describes differentiations in terms of multi-echelon inventory models and multi-objective concepts (service levels, cost optimization, batch and emergency deliveries etc.). Current trends and developments such as sharing economy and customer integration concepts as well as cooperation and new technologies are elaborated in Sect. 14.5 (state-of-the-art). Future topics for distribution management research are discussed briefly and further reading materials are listed in Sect. 14.6.

M. Klumpp (B) FOM University of Applied Sciences and University of Duisburg-Essen, Essen, Nordrhein-Westfalen, Germany e-mail: [email protected] S. Heragu Industrial Engineering and Management, Oklahoma State University, Stillwater, OK, USA e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_14

305

306

M. Klumpp and S. Heragu

Fig. 14.1 Connectivity of marketing and logistics in distribution

14.1 Definition and Objectives in Outbound Logistics and Distribution Management Distribution entails providing the customer of a value chain with the desired quantity, quality, and type of products specified in a customer order. Therefore, this function in the operations, logistics, and supply chain domain is of high importance as customer satisfaction, quality, service appeal, and brand quality are finally determined here. The definition of outbound logistics and distribution management is connected to the “place” function in the strategic marketing concept (“4 P”: Product, Price, Promotion, Place). Figure 14.1 outlines the interconnection of various fields inbetween marketing and logistics, especially distribution. Therefore, distribution is per se customer-centered and relates to the requirements and satisfaction of customers. Customers themselves can be different entities, from internal corporate customers to external corporate customers (“Business-toBusiness—B2B”) or external private customers (“Business-to-Consumer—B2C”). One exemplary definition for distribution management is1 : • The planning, implementation, evaluation, and control of the physical movement of goods from a supplier or manufacturer to the point of sale and the customer, the connected information and financial flows as well as return flows of goods. This includes the management of resources and processes used to deliver a product from a production location to the point-of-sale, including storage at warehousing locations, delivery to retail distribution points or customer sites or homes. 1 Compiled by the authors based on different descriptions, see for alternative explanations for exam-

ple: Blanchard (2007), pp. 121ff); Hugos and Thomas (2006, pp. 34ff).

14 Outbound Logistics and Distribution Management

307

Distribution management also includes the determination of optimal quantities of a product for delivery to specified warehouses or points-of-sale in order to achieve the most efficient, sustainable, transparent, and satisfactory delivery to customers. Distribution management is therefore an overarching, strategic concept that refers to a series of activities and processes such as packaging, inventory, warehousing, and transportation. Usually it involves establishing the following systems and processes among others: • A packaging process with the connected packaging system to ensure secure shipping (avoiding damages to the goods as well as the environment). The reader is encouraged to read Chap. 13, which discusses packaging logistics. • A set of distribution channels to distribute the goods directly from the manufacturer, or a manufacturer’s warehouse, a third-party warehouse, or a third party who matches customer demand with manufacturing supply. Some of the questions that need to be addressed include: How many manufacturing plants and warehouses to locate? Where to locate them? How large should each be? What items should be produced or stocked at each location? What are the levels of production or inventory at each location? Models and algorithms are usually employed for addressing these problems. • A transport system to move the goods to multiple geographical areas. • A tracking system that ensures the right goods reach the right place at the right time in the right quantity. • A system to return goods from retail partners and customers (reverse logistics). The reader is also encouraged to read Chap. 16 on closed loop supply chains. • A continuous evaluation and tracking of places and retail channels where the product can be placed such that there is a maximum opportunity to distribute it to the customer (B2C). Effectively managing the entire distribution process is critical to financial success and corporate survival. The larger a corporation or the greater the number of supply points a company has, the more it will need to rely on elaborate concepts and automation to effectively manage the distribution process. Furthermore, typical objectives within the field of outbound logistics and distribution management are: • Quality: Quality from the customer perspective is an important objective as total customer satisfaction may largely depend on distribution quality (“place”). It may, for example, include the product being free from damages and according to the customer specifications. Quality preservation may also require strong consolidation (e.g., for fresh products) but also speed (e.g., the cut flowers distributed by Flora Holland (largest flower auction in the world lose some 10% of their value every day). • On Time and In Full (“OTIF”): This key performance indicator (KPI) is measured by many companies in their inbound/supply section. It requires the ordered product to be delivered at the specified time according to the order requirements (quantity, packaging, and other requirements). In addition, private customers (“B2C”) are

308









M. Klumpp and S. Heragu

increasingly aware of this key service element. For example, a customer can select his or her parcel in e-commerce with specified delivery windows.2 Customer Satisfaction and Transparency: Tracking and Tracing are important service features as well as a friendly and helpful support (phone/mail/chat hotline or other channels). Transparency in transportation cannot be rated high enough for its influence on total customer satisfaction. Flexibility/Stability/Scalability/Resilience: The implemented processes as well as systems within the distribution function have to be flexible (be able to accommodate changes in volume, destination, and product requirements) and stable given possible external shocks from weather incidents, strikes or other impacts. Furthermore, scalability, i.e., the ability to significantly increase or decrease processed and transported cargo volumes is required, especially in e-commerce environments where online orders of a product can skyrocket within minutes or hours (e.g., with a new book release, a new film or any social media news for example on fashion items). Finally, resilience as a further requirement also falls into this category, to ensure that the distribution system is able to withstand long-term as well as short-term changes. Examples include road toll introduction, emission legislation on vehicles, severe customer and market shifts or the impact of political events like embargos or even civil wars. Cost-Efficiency: The entire outbound and distribution system, including the other corporate and supply chain processes, must be cost-efficient. This implies the use of state-of-the-art technologies and automation, economies of scale, effective personnel and human resource (HR) management3 as well as rigorous selection and cooperation processes for cost management amongst all distribution partners and suppliers as well as logistics service providers. Sustainability and CSR: Finally, societies and customers today also rightfully expect sustainable distribution solutions with a low carbon emission impact and reduction in noise, congestion, and smog. Thus, minimizing the ecological footprint and adhering to corporate citizenship as well as social responsibility principles are important considerations in outbound and distribution systems.

It must be recognized that the deliver “function” of distribution is repeated several times throughout the supply chain as depicted (see the two alternative versions of the SCOR model in Fig. 14.2). Thus, although we discuss the distribution field primarily from the point of view of the end customer in a supply chain, most concepts can be used also at other places and processes throughout the network of supply chain operations as depicted in Fig. 14.2. For example, the deliver function occurs several times throughout typical supply chain networks and therefore has multiple places throughout the supply chain operation.

2 See for example the DHL (http://www.dhl.de/en/paket/pakete-empfangen/paketankuendigung-w unschtag.html) or UPS solutions for private B2C customers https://www.ups.com/content/us/en/b ussol/browse/personal/delivery_options/my_choice.html. 3 See Chap. 10 on Human Resource and Knowledge Management.

14 Outbound Logistics and Distribution Management

309

Fig. 14.2 SCOR model of supply chains (Supply Chain Council 2006, p. 3/22)

14.2 Case Study: Ann Inc. Ann Inc., was founded in 1954 as Ann Taylor. It offers two brands—Ann Taylor and Loft, and is a retailer specializing in women’s apparel, shoes and accessories. It operates over 1,000 retail stores in most states in the US and Canada. To stock and restock its retail stores, Ann Inc., has one warehouse located in Louisville, Kentucky (USA). The 256,000 square feet Distribution Center (DC) receives supplies from all over the world. More than 60% of the inventory is cross-docked. In other words, these items come into the DC from a factory in the Far East via container ships, rail and road, already packaged and labeled to be dispatched to one of the 1,000 retail stores. They are unloaded from the inbound truck, go through a sortation system and are loaded to a shipping truck without the item spending any time on a shelf in the DC. Some

310

M. Klumpp and S. Heragu

items such as a carton of T-shirts do, but wedding gowns come in from the factory already in a clothes hanger, go through a sortation system and are transported through the supply chain all the way to a retail store on a hanger. Some basic design and operational questions that arise in outbound distribution and logistics relative to a DC include: • What percentage of the goods handled should be cross-docked? • What material handling systems should be installed for sorting and moving the goods from inbound trucks to outbound trucks? • How many part-time workers to assign in each month of the year? Ann Inc., decided to locate its warehouse in Louisville, KY so that it could utilize UPS’ extensive distribution network (with its only air-hub located in Louisville) and be able to ship packages to stores overnight. Because the retail stores are located in fashionable shopping districts where the cost to rent or lease retail space is relatively high, these stores wish to use more of their available space for item display than to maintain store inventory. As a result, these stores require replenishment of their inventory two or three times a week. It is the primary function of the DC to ensure the required items are delivered to each of the 1,000 stores as requested on time even though the factories may be located in multiple countries all over the world. Because of the air-hub location, Ann Inc., can ship as late as 10 pm on a given day and be able to deliver to any of its US stores the very next day. Because storing items in a warehouse is a non-value added activity, Ann Inc., also prefers to minimize the time cartons containing high-value items spend time on storage racks and thus opts to cross-dock most—more than 60%—of its items even though they may be coming from as far away as China. Like most retailers, Ann Inc., faces significant demand fluctuation. Its DC activity tends to peak in the October–December time period. Thus, to keep labor costs down, it relies on seasonal employees to fulfill the orders. In order to ensure the availability of seasonal employees, Ann Inc., offers competitive wages as well as annual healthcare and retirement benefits. Of course, if an employee wants full-time employment, Ann Inc., will provide that to that employee, but he or she must be willing to work in all areas of the DC, not just in order picking in order to provide a reasonable level of flexibility. As can be seen from the above example, distribution logistics covers a myriad of design and operational problems and each must be solved efficiently for the organization to succeed as a whole.

14 Outbound Logistics and Distribution Management

311

14.3 Core Concepts of Distribution (Basic) 14.3.1 Conceptual Definitions and Levels of Distribution Management On a strategic (“macro”) level—directly derived from the definition of distribution management and outbound logistics—a balance between four interlinked areas of distribution decisions must be taken. As depicted in Fig. 14.3, the geographical reach and the markets where products must be distributed have to be declared (in line with marketing strategies and concepts, e.g., if products shall be available and be delivered to South Africa). Next, the level and details of delivery service in the distribution domain must be determined. Some examples are listed below: • delivery time(s)/standard, express and exceptional defined as the proposed time, • customer communication and services, e.g., order acknowledgement, track and trace services of the delivery, • standard packaging definitions to avoid damaged goods as well as exceptional packaging options, • request and feedback services and times. Third, elements of differentiation must be elaborated, e.g., how to differentiate in distribution from competitors. Finally, an efficiency and rationalization component needs to be outlined in order to keep the distribution cost levels in balance (e.g., where to use automated communication and transport applications). On a tactical (“meso”) level, additional decisions must be taken regarding the warehousing structure and points of distribution (see Fig. 14.4). In this case, interconnected determinations of the vertical structure (“How many levels of warehouses and distribution points?”) as well as the horizontal structure (“How many warehouses per level, at what locations, with what specific functionalities, e.g., return processes?”) of distribution must be determined. Fourth, cooperation and distribution partners are selected and managed and fifth, transport modes and vehicles must be defined (e.g., selection of modes of transportation based on cost, time, and other considerations). Again, it must be stressed that all decision areas are interlinked, meaning that decision in one field will have an impact on other decision areas.

Reach / Markets

Delivery Service

Differen a on

Ra onalisa on

Fig. 14.3 Strategic (“Macro”) decision fields in distribution

312

M. Klumpp and S. Heragu

VerƟcal: Central vs. Decentral

Horizontal: Loca ons, Number, Func onali es

Coopera ons / Distribu on Partners

Transport Modes and Vehicles

Fig. 14.4 Tactical (“Meso”) decision fields in distribution

Fig. 14.5 Operational (“Micro”) decisions and setup in distribution

On an operational (“micro”) level, additional decisions must be taken regarding the specific setup and capacities of distribution points and warehouses. As depicted in Fig. 14.5, the structure of such distribution systems may be very complex when the product portfolios are large. Note that a typical retailer offers more than 10.000 stock keeping units (SKUs). As the number of warehouses increases, the operating, inventory, and transportation costs curve changes as shown in Fig. 14.6. The planner thus needs to trade-off the various costs involved and determine the optimal number of warehouses to operate.

14 Outbound Logistics and Distribution Management

313

Fig. 14.6 Cost optimization problem in distribution

14.3.2 Additional Concepts in Outbound and Distribution An important characteristic of a supply chain in general and especially “downstream” in the distribution phase, is the bullwhip effect (see Lee et al. 1997; Carranza Torres and Villegas Morán 2006). This recognizes excess increases of order volumes “upstream” due to information and transparency gaps between the supply chain actors. So, as depicted in Fig. 14.7, if the customer (or a group of customers) orders an increased level of product volumes at the retail point of sale (say in London, UK), usually without further information any (human) actor is prone to adding a security level for good measure in her/his own order volume with her/his supplier. This occurs several times throughout the chain, so that finally order volumes become disproportionately high at the most upstream producer and supplier entities (e.g., in Poland in the example in Fig. 14.7). This empirical recognition of excess logistics costs throughout the supply chain based on limited information and human behavior is utilized in a simulation and education game.4 An elaborate concept for cooperative distribution stemming from the retail sector is the Efficient Consumer Response (ECR) concept. This concept implies the strategic cooperation of producer and retailer in a consumer product area such as food, cosmetics or other fast-selling items; this is enabled by the increase of transparency by data sharing among supply chain partners (including the producer, suppliers and the retailer as well as logistics service providers). The key elements for this retail 4 Compare

for example http://supplychain.mit.edu/supply-chain-games/beer-game or alternative simulations with other universities, e.g.: http://www.beergame.org or http://www.beergame.lim. ethz.ch.

314

M. Klumpp and S. Heragu Customer order spike

Retailer

Retailer increases order volume Warehouse

Warehouse increases order volume Manufacturer

Manufacturer increases order volume

Supplier

t Fig. 14.7 Bullwhip effect

Efficient Consumer Response (ECR) Supply Side – SCM Efficient Replenishment XML

Efficient Operating Standards

Demand Side – CM Efficient Store Assortment

Efficient Promotion

Efficient Product Introduction

Collaborative Planning, Forecasting and Replenishment (CPFR) EDI

EAN

Efficient Administration

Scanning

Enabling Technologies

ILN WWS

SMTP

RFID HTTP

Fig. 14.8 Efficient consumer response concept

logistics concept—implemented and successfully adapted throughout many industries today—are depicted in Fig. 14.8. On the Supply Side/Supply Chain Management (SCM) part of the ECR concept, there are three specific instruments to be used: • Efficient Replenishment aims at furthering the quality and efficiency of all replenishment processes by using cross-dock concepts, the Continuous Replenishment Program (CRP) or Roll Cage Sequencing. CRP has the basic idea that long- and mid-term forecasts are shared among all the partners in a supply chain, enabling adaptation and alignment along the way. This could lead to a significant reduction of frictions and distorted or antagonistic forecast and replenishment decisions between supplier, retailer, logistics service provider or other entities in a supply chain.

14 Outbound Logistics and Distribution Management

315

• Furthermore, Efficient Operating Standards provide the option to use standardized sizes in order to optimize vehicle and storage use. For example in Europe, the cargo heights within trucks and warehouses are standardized for the EUL 1 (1.20 m) and EUL 2 (2.40 m) categories. This allows more efficient transport and storage of cargo, especially when all the partners in a supply chain adhere to these standards. Similarly, a product “base footprint” of 40 by 60 cm is defined in accordance with the European pallet size (120 cm × 80 cm), indicating that four base units can efficiently be put on one pallet. • Finally, Efficient Administration provides the guidelines for further improvements in the management of the supply processes in the (retail) supply chain, e.g., by using resource pooling systems where different partners within the supply chain jointly use resources such as warehouses or transport vehicles. On the Demand Side/Category Management (CM) part of the ECR concept, there are at least four specific instruments to be applied: • The method of Category Management is structured within the realm of marketing addressing the question of how to organize the large portfolio of items offered to customers. To manage this, specific product categories are defined (e.g., the health and cosmetics section, the food category or the clothes category). This allows for a further use in structuring demand and supply and presenting items to customers. • The Efficient Assortment concept elaborates on the preceding categories. The optimal mix of products and quantities offered in the retail system and stores are decided jointly by supplier, retailer, and logistics service partner, taking into consideration customer preferences, production, and transportation constraints as well as other factors. • Within product development and Efficient Product Introduction, again the core value and principle of cooperation and coordination are assumed to be the basis for an efficient introduction process (“ramp-up”) for new products on the retail shelves. • In a similar manner, Efficient Promotions are guided by the principle that specific sales activities should be planned and executed jointly by all the supply chain partners. For example, production and delivery of special promotion products and advertising material within retail stores must be aligned with marketing and PR activities within the retail system. There exist several specific instruments pertaining to the basic enabling technologies for ECR. Examples include: • Electronic Data Interchange (EDI) systems and standards allow the supply chain partners to automatically share and distribute information on sales, stock and production volumes for an efficient alignment of downstream activities such as replenishment orders or production lot decisions. • Additional instruments may include, systems and hardware standards, joint value evaluation and customer research or joint e-commerce initiatives of all supply chain partners.

316

M. Klumpp and S. Heragu

These instruments within ECR are feasible and require the unrestricted cooperation of supply chain partners, especially on information exchange and transparency. Finally, there are many quantitative outbound distribution problems in practice that involve transporting goods from a warehouse to a customer or from a manufacturing plant to a warehouse. In these problems, the objective is to minimize the cost of transporting the goods from their source(s) to their destinations, while satisfying demand and supply constraints (Fig. 14.9). A transportation model is typically applied in order to model such problems—an example is shown below. Consider the following notation: Parameters: m n cij Si Dj

number of plants number of warehouses cost of transporting a unit from plant i to warehouse j supply capacity of plant i (in units) demand at warehouse j (in units) Decision Variable:

x ij number of units transported from plant i to warehouse j The transportation model is:

S1

1 c13

S2 2 S3 3

c11 c21

c12

1 D1 c22

c23

c32

c31

c33 c2n

cm2

cm3

2

D2

3

D3

cm1 c1n

Sm m

cmn

Fig. 14.9 Graphical representation of a transportation problem

n

Dn

14 Outbound Logistics and Distribution Management

Model

m  n 

317

ci j xi j

(14.1)

xi j ≤ Si i  1, 2, . . . , m

(14.2)

i1 j1

Subject to

n  j1

m 

xi j ≥ D j

j  1, 2, . . . , n

(14.3)

i1

xi j ≥ 0 i  1, 2, . . . , m; j  1, 2, . . . , n

(14.4)

The objective function (14.1) minimizes the total cost of transporting the goods from the plants to the warehouses. Constraint (14.2) ensures that the total number of units leaving each plant does not exceed its capacity. Constraint (14.3) ensures that the total number of units received at each warehouse is greater than or equal to the demand at that location. If the total demand equals the total supply, we have a balanced transportation problem. If the total demand exceeds the total supply (or the total supply exceeds the total demand), we can add a dummy plant (warehouse) that produces (absorbs) the excess demand (capacity) to convert the unbalanced problem to a balanced transportation problem. In the balanced transportation problem, all the inequality signs [ < in constraint (14.2) and > in constraint (14.3)] can be replaced by equality signs. Constraint (14.4) ensures that the number of units transported from each plant to each warehouse are nonnegative. Although integer restrictions are not imposed on the x ij variables, it should be noted that these are automatically satisfied in the optimal solution provided the S i and Dj values are all integers.

14.4 Multi-perspective Outbound Management (Advanced) 14.4.1 Location Analysis for Distribution Embedded in outbound and distribution questions and concepts are usually a series of operational questions demanding a multi-perspective analysis and decision process. For example location decisions must be made. As seen in the case study in Sect. 14.2, these decisions involve estimating a series of cost and quality outcomes. Thus, location decisions must be made carefully. In the German mail distribution system example depicted in Fig. 14.10, a series of warehouse and distribution locations for the delivery of parcels and letters is necessary so that the customers are served with speed and flexibility, while keeping costs low, and operating in a sustainable manner by minimizing the pollution from transportation. Such locations are re-evaluated regularly, see for example Fig. 14.10. For the mathematical analysis of transportation and location problems, the “Steiner-Weber-Problem” is formulated as follows—minimizing the transportation

318

M. Klumpp and S. Heragu

Fig. 14.10 Location map for Germany/Mail System (DPDHL)

costs depending on the location decisions (x and y coordinates in a geographical setting. Note the similarity to the basic transportation model introduced in the previous section): Model: min

m  n  i1 j1

with the restrictions:

ci j ωi j



xi − u j

2

 2 + yi − v j

14 Outbound Logistics and Distribution Management m 

ωi j ≥ D j

319

j  1, . . . , n

i1 n 

ωi j ≤ Si i  1, . . . , m

j1

ωi j ≥ 0 i  1, . . . , m; j  1, . . . , n Parameters: m n x/y u/v cij Dj Si

number of originating locations number of destination locations coordinates of origin locations coordinates of destination locations cost of transporting a cargo unit per distance unit demand at destination j (in units) supply capacity of origin i (in units)

Decision Variable: ωij number of units transported from origin i to destination j

14.4.2 Comprehensive Location-Allocation Model In this section, we present a comprehensive production-distribution model that considers a real-world problem (see Geoffrion and Graves 1974). Multiple product types are produced at several plants with known production capacities. The demand for each product type at each of several customer areas is also known. The products are shipped from plants to customer areas via intermediate warehouses with additional restrictions, e.g., each customer area is serviced by only one warehouse to improve customer service. Upper and lower bounds on the capacity of each warehouse, potential locations for these, inbound and outbound transportation costs at each of these warehouses (i.e., from each plant and to each customer area), and the fixed cost of opening and operating a warehouse at each potential location are known. The problem is to find the optimal locations for the warehouses, the corresponding capacities, the customers served by each warehouse, and how products are to be shipped from each plant, in order to minimize the fixed and variable costs of opening and operating warehouses as well as the distribution costs. Consider the following notation: S ij Dil Fk

production capacity of product i at plant j demand for product i at customer zone l fixed cost of operating warehouse k

320

M. Klumpp and S. Heragu

V ik cijkl

unit variable cost of handling product i at warehouse k average unit cost of producing and transporting product i from plant j via warehouse k to customer area l Upper bound on the capacity of warehouse k Lower bound on the capacity of warehouse k number of units of product i transported from plant j via warehouse k to customer area l  1 if warehouse k serves customer area l ykl  otherwise

UC k LC k x ijkl

 zk 

Minimize

q p r  s   

1 if warehouse is opened at location k otherwise

ci jkl xi jkl +

i1 j1 k1 l1

Subject to

r  s 

p s   i1 l1

Dil

r 

Vik ykl +

k1

r 

Fk z k

(14.5)

k1

xi jkl ≤ Si j i  1, 2, . . . , p; j  1, 2, . . . , q

(14.6)

k1 l1 q 

xi jkl ≥ Dil ykl i  1, 2, . . . , p; k  1, 2, . . . , r ; l  1, 2, . . . , s

(14.7)

j1 r 

ykl  1 l  1, 2, . . . , s

(14.8)

k1 p s   i1 l1 p s  

Dil ykl ≥ LCk z k k  1, 2, . . . , r

(14.9)

Dil ykl ≤ U Ck z k k  1, 2, . . . , r

(14.10)

i1 l1

xi jkl ≥ 0 i  1, 2, . . . , p; j  1, 2, . . . , q; k  1, 2, . . . , r ; l  1, 2, . . . , s (14.11) ykl , z k  0 or 1 k  1, 2, . . . , r ; l  1, 2, . . . , s

(14.12)

The objective function (14.5) minimizes the total cost of producing goods at the plants and transporting them to customers via warehouses. It also minimizes the fixed and variable costs of opening and operating the required number of warehouses. Constraint (14.6) ensures that the capacity constraints for making each product at each plant are not violated. Constraint (14.7) ensures that the demand of each product at each customer zone is met. Constraint (14.8) requires that each customer area is serviced by a single warehouse. Constraints (14.9) and (14.10) have a dual purpose.

14 Outbound Logistics and Distribution Management

321

Not only do they enforce the upper and lower bound on the warehouse capacity, they also “connect” the ykl and zk variables. Because a warehouse can serve a customer area only if it is opened, we must have ykl  0 for l  1, 2, . . . , s if zk  0. If on the other hand a particular zk  1 for some k, then ykl  1 for the same k and at least one l ∈ {1, 2, . . . , s }. Constraints (14.11) and (14.12) are the usual nonnegativity and integer value constraints. We can add more linear constraints not involving x ijkl variables to the above model to: • Impose upper and lower limit on the number of warehouses that can be opened; • Enforce precedence relations among warehouses (e.g., open warehouse at location 1 only if another is opened at location 3); and • Enforce service constraints (e.g., if it is decided to open a certain warehouse, then a specific customer area must be served by it). Other constraints that can be added are discussed further in Geoffrion and Graves (1974). Such constraints reduce the solution space, and for this model, they allow a faster solution while giving the modeler much flexibility. Real-world problems such as the Hunt-Wesson Foods, Inc. location-allocation problem considered in Geoffrion and Graves (1974), which had more than 11,000 constraints, 23,000 x ijkl variables as well as 700 ykl and x k binary variables, have been rather easily solved using a modified version of Benders’ decomposition algorithm.

14.4.3 Transport Mode Analysis and Selection A further complex decision within distribution systems usually is the question of transport mode as well as vehicle selection and fleet management. For many transportation tasks within distribution, the road is the premier mode of choice due to its flexibility. In fact, road travel is typically the only choice for the last mile delivery. However, when transporting via long distances or transportation, speed is important. Thus, other modes of transportation such as rail, ship, or air are utilized. Besides speed and cost, other considerations such as sustainability, transparency concerns, service supplier market power or domination, and other must be incorporated. For example, several companies such as Volkswagen or Mercedes are taking over distribution tasks from logistics companies due to the lack of quality or trust. Similar patterns occur in the other direction as well where logistics service providers operate distribution warehouses for their customers (e.g., logistics providers for Aldi and Samsung in Germany also manage their inventory). Typically, transport mode analysis and decisions are derived from a set of distinction categories such as cost, distance, sustainability (carbon emission comparison by mode), speed or other specifics such as transport density as with the following example of a transport mode analysis for personnel transportation in an urban area (Fig. 14.11).

322

M. Klumpp and S. Heragu

Fig. 14.11 Transport mode comparison (example passenger transport)

14.4.4 Vehicle Routing in Distribution In this section, we present a model that is often used for outbound distribution management. It is used for last-mile delivery. Consider the problem faced by a parcel delivery company in its hub location. It wants to determine the number of vehicles required to: (1) serve its customers (pick-up or deliver parcels) so that each customer is visited once and only once per day, (2) the vehicle capacity is not exceeded, and (3) the total travel time is minimized. Using the notation provided next, we present the model in Chandran and Raghavan (2008) for the capacitated vehicle routing problem (VRP). Parameters: Si T ij Di C

shortest time to travel from customer i to the depot, i  1,2,…, n shortest time to travel from customer i to customer j, i, j = 1,2,…, n demand at customer i, i = 1,2,…, n capacity of each vehicle Decision Variables:  1 if customer j is immediately visited after customer i in a tour xi j  0 otherwise  yi j 

1 if customers i and j are visited by the same vehicle 0 otherwise

14 Outbound Logistics and Distribution Management

 pi   qi 

323

1 if node i is the first customer visited in a tour 0 otherwise 1 if node i is the last customer visited in a tour 0 otherwise

The VRP model is: Minimi ze

n 

Si ( pi + qi ) +

i1 n 

Subject to Di +

n−1  n 

Ti j xi j

(14.13)

i1 ji+1

D j yi j ≤ C, i  1, 2, . . . , n

(14.14)

ji+1

yi j + y jk − yik ≤ 1, i, j, k  1, 2, . . . , n : i < j < k xi j − yi j ≤ 0, i, j  1, 2, . . . , n 

(14.15) (14.16)

j−1

pj + qi +

xi j  1,

i1 n 

j  1, 2, . . . , n

(14.17)

xi j  1, i  1, 2, . . . , n

(14.18)

ji+1

pi , qi , xi j , yi j  0 or 1, i, j  1, 2, . . . , n : i < j

(14.19)

The objective function (14.13) minimizes the total travel time to and from the depot as well as between the customers. Constraint (14.14) ensures that the sum of the demands at customers served by a vehicle does not exceed its available capacity. Constraint (14.15) ensures that if customers i and j as well as customers j and k are served by a truck, then, customers i and k must also be served by the same truck. Constraint (14.16) ensures that customer j can be visited immediately after visiting customer i, only if both customers are served by the same vehicle. Constraints (14.17) and (14.18) ensure that each customer is served by exactly one truck. Constraint (14.19) imposes (binary) integer restrictions on the decision variables. Numerous heuristics are available to solve the VRP. See Bodin et al. (1983) for a survey of the various algorithms and models for this problem. Alfa et al. (1991) have proposed a combined 3-opt and simulated annealing algorithm for this problem. We will not provide any algorithm here, but recommend the reader to consult the above two sources for ways of solving the VRP.

324

M. Klumpp and S. Heragu

Before Cooperation

Fig. 14.12 Cooperative distribution for Urban areas

14.4.5 Last Mile and Urban Distribution The last leg in distribution is usually known as the “last mile” delivery. This is critical for many highly populated urban areas, because congestion, availability as well as transport speed and flexibility and other factors must be considered. Cooperation among competitors (sharing transportation services or warehouse space among competitors) is one way to cope with these problems. Instead of travelling independently into urban areas as depicted in Fig. 14.12, cooperation among partners, consolidating shipments outside the city area and implementing an optimized tour plan within the city with fewer vehicles allows competitors to share resources and minimize environmental impact while meeting their own delivery requirements efficiently.

14.5 New Developments in Distribution (State-of -the-Art) There are many developments taking place in logistics and distribution. Some of the innovative concepts are outlined in this section. Third-party logistics is an increasingly important and successful business model for many companies. Consider United Parcel Services (UPS). Not only is UPS a distributor of goods, but it also manages inventory and provides value-added services for its customers. A few years ago, UPS was not only responsible for transporting Toshiba notebooks, but it also served as a repair facility. A customer who had to have his or her laptop repaired simply had to go to Toshiba’s website, print out an address label, and drop it off for pick-up by UPS. UPS employees would then transport the laptop to their Louisville air-transportation hub, where a set of trained technicians (also UPS employees) would then repair the laptop and by the end of the day put the laptop back into the distribution network for delivery to the customer the next day. The same is true for many other modern-day logistics service providers such as FedEx, DHL, Schenker, Excel and others.

14 Outbound Logistics and Distribution Management

325

The Unilever warehouse in Beauvais, France, for example, has a fully automated warehouse in which DHL drivers load and unload pallets into an automated high rise warehouse. Autonomous vehicles (see Cai et al. 2014) then pick up these pallets, travel along guided rails and elevators to deposit pallets into their designated locations (see the red vehicles visible in the background in Fig. 14.13 and more clearly in Fig. 14.14). Once stored, the pallets are called as needed for order-picking in another part of the warehouse where robots put together customer orders also in a fully automated manner. Many forms of automation have been employed in the industry. Some technologies (e.g., the Kiva or Grey Orange robots) bring pods of items to the operator for picking, whereas others, e.g., the Savoye logistics system described in Chap. 15 as well as the Symbotic and the Swisslog mobile robots bring totes to the order picker. See https://www.therobotreport.com/news/goods-to-man-robotic-syst ems for additional details. Amazon, for example, is experimenting with the delivery of individual orders directly to the customer using drones or unmanned aerial vehicles. While there are operational and regulatory hurdles that need to be cleared, this appears to be an important technology that can help with the last-mile delivery. Distributors find that they can manage long-haul distributions effectively using air, rail, road, and river systems, but the last-mile delivery is highly inefficient for a variety of factors including traffic, congestion of road networks, and involvement of humans in the last-mile delivery—not only truck drivers, but also customers who receive the package. For that reason, last-mile distribution is often considered the most expensive part of a distribution system. As a result, companies such as DHL are now experimenting using driverless vehicles or even robots for the last-mile delivery.

Fig. 14.13 An automated warehouse system (Savoye Logistics)

326

M. Klumpp and S. Heragu

Fig. 14.14 Automated warehouse system using autonomous vehicles and lifts (Savoye Logistics)

A few of the many emerging topics in logistics and transportation are briefly mentioned below: • New collaboration forms and sharing economy (buddy transport systems): Many trailers and containers travel fully or partially loaded one way and empty or almost empty in the reverse direction. By sharing information, it is possible to reduce empty or partially loaded travel. • New functions and roles of 3PL/4PL are expected, e.g., in the essential areas of logistics cooperation and pooling of resources (warehousing) as well as optimization (see Wu et al. 2016). In this field, AI applications are also expected to have a major impact (also see Chaps. 24 and 28). • Peer to peer-concepts are predicted to emerge and affect many transportation segments, for example by customer integration in delivery processes within the CEP sector (e.g., Uber). • For distribution purposes on a global scale, we also expect significant regional changes in demand and volume patterns, e.g., the continuing rise of Intra-Asian transportation (e.g. from/to India), increasing importance of Africa and the Mid-

14 Outbound Logistics and Distribution Management

327

dle East for transportation as well as the impact of future trade agreements on transportation volumes (CETA/TTIP). • Automated transportation will play a role especially in specific niche segments (delivery robots, drones); in Germany, for example Deutsche Post DHL is testing delivery robots (“PostBOT”) supporting mail delivery in urban areas.5 • Finally, there will be a prominent role for predictive analytics applications, especially for outbound distribution.

14.6 Outlook and Further Reading In 2014, MHI convened a set of brainstorming sessions in four cities in the US (Atlanta, Chicago, Los Angeles, and Washington, D.C.) with more than 100 thought leaders from academia, industry, and government.6 The charge to this group was to predict logistics and distribution challenges faced by the materials handling and supply chain industry in the year 2025 and to identify the technologies that would need to be developed in order to overcome these challenges. The deliberations of these groups led to the development of a roadmap. The societal, economic, and technological trends identified in the roadmap that will affect the state of the logistics and distribution in 2025 are: • • • • • • • • • •

E-commerce growth, Competition, Mass personalization, Urbanization, Mobile and wearable computing, Robotics and automation, Sensors and the Internet of Things, Big data and predictive analytics, The changing workforce, Sustainability.

A number of the above topics have been briefly discussed in this chapter. We encourage the reader to consult the MHI roadmap for additional discussions on these topics. One of the topics that is mentioned as an addendum is 3D printing. Along with driverless vehicles (ground-based or aerial) and virtual reality, this technology will have significant impact on the logistics and distribution industry. There will be reduced need for transportation of small products and replacement parts, because these can now be made on a 3D printer. 3D printing also makes mass customization a reality. Chapter 23 discusses 3D printing and its impact on the supply chain and the reader is encouraged to review that chapter. Although many companies claim they 5 See:

http://www.dpdhl.com/en/media_relations/press_releases/2017/new_delivery_robot_suppor ts_mailmen.html. 6 Compare: http://www.mhlroadmap.org.

328

M. Klumpp and S. Heragu

Fig. 14.15 Logistics trend radar (DHL 2016)

are able to provide customized products to individual consumers, they only allow the consumer to select from a list of options, which when combined makes the product somewhat unique. On the other hand, 3D printing truly allows the customer to customize products to suit their physical attributes, taste, style, aesthetical requirements, and others. This will further revolutionize business, markets and value chains throughout all industries (Fig. 14.15). Similar studies have been implemented for logistics in general, e.g., in Europe by DHL such as the “Delivering Tomorrow” scenario study (DHL 2012) or the “Logistics Trend Radar” (several editions, latest 2016, DHL 2016). The above figure outlines such a trend radar picture with an overview of relevant technology and society chances for logistics developments. For further reading we also refer the reader to the following sources: For more information on distribution in urban areas, see Antun (2016), Kant et al. (2016), Sakai et al. (2016) or Klumpp et al. (2014); To address automation developments in distribution, see for example, Arendt et al. (2016) or Klumpp et al. (2013b); For elaborate concepts in routing, we refer to Simchi-Levi et al. (2015); For advances in distribution center management, see Cipres et al. (2014); To explore supply chain and distribution risks in detail, see Klumpp and Abidi (2013) or Waters (2007);

14 Outbound Logistics and Distribution Management

329

Finally, for general trends and developments, refer to Speranza (2016), Novaes et al. (2014) or Klumpp et al. (2013a).

References Alfa AS, Heragu SS, Chen MY (1991) A 3-OPT based simulated annealing algorithm for vehicle routing problems. Comp & Ind Eng 21(1–4): 635–639 Antun JP (2016) Corporative trends in Urban distribution of goods in Mexico City. Transp Res Procedia 18:51–58 Arendt F, Klein O, Barwig K (2016) Intelligent control of freight services on the basis of autonomous multi-agent transport coordination. In: Zijm H et al (eds) Logistics and supply chain innovation. Springer, pp 313–324 Blanchard D (2007) Supply chain management best practices. Wiley, Hoboken Bodin L, Golden B, Assad A, Ball M. (1983) Routing and scheduling of vehicles and Crews—the state of the art. Comput Oper Res 10(2):63–211 Cai X, Heragu SS, Liu Y (2014) Modeling and evaluating the AVS/RS with tier-to-tier vehicles using semi-open queuing network. IISE Trans 46(9):905–927 Carranza Torres OA, Villegas Morán FA (2006) The Bullwhip effect in supply chains—a review of methods, components and cases. Palgrave Macmillan, Houndmills, New York Chandran B, Raghavan S (2008) Modeling and solving the capacitated vehicle routing problem on trees. In: Golden BL, Raghavan S, Wasil EA (eds) The vehicle routing problem: latest advances and new challenges. Springer, pp 239–262 Cipres D, Millan C, Errasti A, Larrode E (2014) Integrated system for combining decisional problems in a distribution centre. In: Clausen U, ten Hompel M, Meier JF (eds) Efficiency and innovation in logistics. Springer, pp 91–108 DHL (2012) Delivering tomorrow, logistics 2050—a scenario study. DHL, Bonn DHL (2016) Logistics trend radar 2016. DHL, Bonn Geoffrion AM, Graves GW (1974) Multicommodity distribution system design by benders decomposition. Manag Sci 20(5):822–844 Hugos M, Thomas C (2006) Supply chain management in the retail industry. Wiley, Hoboken Kant G, Quak H, Peeters R, van Woensel T (2016) Urban freight transportation: challenges, failures and successes. In: Zijm H et al (eds) Logistics and supply chain innovation. Springer, pp 127–140 Klumpp M, Abidi H (2013) Supply chain risk management for macro risks. In: Munson C (ed) The supply chain management case book. Pearson Education Klumpp M, Marner T (2014) Comprehensive Urban road toll evaluation system. Int J Transp Traffic Eng 4(1):14–34 Klumpp M, Clausen U, ten Hompel M (2013a) Logistics research and the logistics world of 2050. In: Clausen U, ten Hompel M, Klumpp M (eds) Efficiency in logistics, lecture notes in logistics. Springer, Heidelberg, pp 1–6 Klumpp M, Kandel C, Wirsing E (2013b) Cargo telematics for operational transport excellence and strategic knowledge management. In: Kreowski H-J, Scholz-Reiter B, Thoben K-D (eds) Dynamics in logistics, lecture notes in logistics series. Springer, Heidelberg, pp 71–84 Klumpp M, Abidi H, Marner T (2014) Qualification hurdles and chances for E-mobility. Int J Transp Traffic Eng 4(3):253–268 Lee HL, Padmanabhan V, Wang S (1997) The Bullwhip effect in supply chains. Sloan Manag Rev 38:93–102 Novaes AGN, Bez ET, Burin PJ (2014) Fault detection in dynamic vehicle routing operations. In: Kreowski H-J, Scholz-Reiter B, Thoben K-D (eds) Dynamics in logistics. Springer, pp 14–34 Sakai T, Kawamura K, Hyodo T (2016) Logistics facility distribution in Tokyo Metropolitan area: experiences and policy lessons. Transp Res Procedia 12:263–277

330

M. Klumpp and S. Heragu

Simchi-Levi D, Chen X, Bramel J (2015) The logic of logistics. Part IV on Vehicle routing. Springer, pp 299–376 Speranza MG (2016) Trends in transportation and logistics. Eur J Oper Res. https://doi.org/10.101 6/j.ejor.2016.08.032 Supply Chain Council (2006) Supply-chain operations reference-model. SCOR overview version 8.0, Supply Chain Council, Washington Waters D (2007) Supply chain risk management—vulnerability and resilience in logistics. Kogan Page, London, Philadelphia Wu Z, Xu J, Xu Z (2016) A multiple attribute group decision making framework for the evaluation of lean practices at logistics distribution centers. Ann Oper Res 247:735–757. https://doi.org/10. 1007/s10479-015-1788-6

Chapter 15

Warehousing Sunderesh Heragu

Abstract In this chapter, topics of interest in the design, analysis, management, and operations of warehouses are discussed. An overview of how warehouses have been in use since the dawn of civilization and how they have changed recently is first provided. Then, the basic elements of a warehouse are discussed and illustrations of the elements of a warehouse, including the materials-handling systems currently used in warehouses are presented. A simple model for determining the warehouse footprint is discussed. A case study involving a consumer goods manufacturer, followed by advanced models for allocating products to a warehouse (and thereby determining the size of the main functional areas in a warehouse), storage policies, corresponding models for the two policies, as well as routing strategies for order pickers are also presented. The chapter is concluded by providing a discussion of state-of-the-art topics in warehousing.

15.1 History of Warehousing It is almost impossible to time the supply of produced goods (supply) and their consumption (demand) so that they occur at the same instant of time. For a variety of factors, including the fact that production and demand consumption typically do not take place at the same location, economies of scale in production, and other factors, goods must be stored either at the source of production, consumption point or at an intermediate location. Thus, warehouses become necessary for manufactured goods. On the other hand, warehouses are not required for services because a service not consumed when it is made available (e.g., an airline seat before a flight takes off), is lost forever. Warehouses have been around in one form of another since the dawn of human civilization. Consider the Harappa and the Mohenjo-Daro civilizations, built along S. Heragu (B) Industrial Engineering and Management, Oklahoma State University, Stillwater, OK, USA e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_15

331

332

S. Heragu

the Indus river valley more than 5,000 years ago. Both were expertly planned communities consisting of orthogonal (Manhattan-like) streets, multi-story dwellings with private and public bathroom facilities, sewer system, and a warehouse-like building, possibly for storing grains. This building had vents for humidity control and lanes for inbound and outbound carts that brought materials into the building. Today, a company like Amazon has more than 100 fulfillment centers with another 150 hubs and sortation centers.1 Using a large network of facilities, including their own and that of third-party manufacturers, Amazon can guarantee delivery within two days of an online order. On the other hand, a company like Ali Baba, which is considered the largest retailer in the world, has no warehouses of its own but instead relies on those of its suppliers to ship goods directly from their warehouses to customers (Hu et al. 2014). In this chapter, the models for the design, control, and analysis of warehousing systems are examined. Illustrations of some of the materials-handling technologies used in warehouses, including some of the newer ones that are automated are also provided.

15.2 Design and Analysis of Warehouses (Basic) In this section, the essential warehouse topics at an elementary level are covered so the user can begin to appreciate the design and operational problems that will arise in warehouses.

15.2.1 Basic Elements of a Warehouse As shown in Fig. 15.1, a warehouse typically consists of an outside shell, storage racks, and materials-handling equipment (Askin and Standridge 1993; Heragu 2016). The exterior (outside shell) of a warehouse is used to protect the warehouse goods from the weather elements and for security. Depending upon the item stored (e.g., refined fuel, lumber, plants), a building or outer shell is not necessary and only fencing on the perimeter of the property is needed. Some warehouses, especially in the US, do not display a company logo or have visible signs because they could be storing sensitive or strategic national stockpile material. For protection against terrorist activities, they tend not to have visible signs. In addition to a focus on the elements of a warehouse, designers must also consider two other parameters—warehouse throughput and storage capacity of the warehouse. In fact, the latter often determine the size of the warehouse building, material handling equipment used as well as storage racks selected. The warehouse throughput indicates how many customer orders can be picked in an hour. The storage capacity indicates 1 http://www.mwpvl.com/html/amazon_com.html.

15 Warehousing

333

Fig. 15.1 Cross-section of a warehouse (Savoye Logistics)

the maximum number of pallets that can be stored in the warehouse. The rate at which items are received in the warehouse depends upon a variety of factors including the specific characteristics of the items received (e.g., size, shape, weight, shelf life), transportation costs, order size and receipt frequency. The rate at which items are shipped from the warehouse depends upon the items in customer orders, order sizes and order frequencies. In an ideal world, the rate at which goods are received must be equal to the rate at which goods are demanded from the warehouse. A crossdocking facility attempts to achieve this by placing pallets from inbound trucks on a conveyor and through a sortation system into outbound trucks. In such a facility, there is very little or no need for storage space. It is not always possible to match the supply and demand precisely and hence temporary storage of goods in a warehouse is necessary. The warehouse size, level of automation, type of storage media used depend upon not only the characteristics of the items and the imbalance in the supply and demand of goods, but also on the desired throughput of the warehouse. The main purpose of the storage racks, as their name indicates, is to store goods on racks. However, depending upon the type of goods stored, and the conditions under which they must be stored, there are a variety of racks available. A few examples are provided in Fig. 15.2. Figure 15.2a illustrates a conventional pallet rack. The flow racks shown in Fig. 15.2b have rollers that allow products to be loaded on one end and retrieved at the other. The gentle slope permits a first-in, first-out retrieval policy. The cantilever racks in Fig. 15.2c allow long rods and pipes to be stored. The mobile racks in Fig. 15.2d are on tracks and can be moved. While they maximize storage density because the racks can be stashed against each other, retrieval of goods may take more time because all the racks in front of an item stored in a middle rack will require all the ones in front of it to be moved to create a floating aisle so the rack under consider-

334

S. Heragu

(c) Cantilever racks (Storax, Inc).

(a) Pallet racks (Storax, Inc).

(b) Flow racks (Storax, Inc).

(d) Mobile racks (Storax, Inc).

Fig. 15.2 Alternate types of storage racks

ation is accessible. There are many other types of racks and the reader is encouraged to consult Heragu (2016) and Tompkins et al. (2010), among other sources.

15.2.2 Storage and Material Handling Systems There is a large variety of materials-handling system used in warehouses. There are many ways of classifying them. In this chapter, a simple classification is used. • Person (or equipment) to item • Item to person (or equipment). Depending upon the extent to which humans are involved in order picking, a warehouse can have no automation, be semi-automated or fully automated. The

15 Warehousing

335

warehouses in the U.S. tend to have little to no automation. Many only have sophisticated forklifts that require operators and/or sortation systems. The European and Japanese warehouses tend to have more automation. This is partly because space is more constrained in Europe and Japan, and thus warehouses have a smaller footprint and more height (e.g., 10–30 m). US warehouses on the other hand have wider footprints and smaller heights (less than 10 m). Taller warehouses will require automated equipment for safe order picking. Automation also yields benefits such as improved accuracy, less damage to goods, and efficient operations. However, their capital costs are high. Figure 15.3 illustrates several manual, semi-automated, and fully automated warehouses. The forklift shown in Fig. 15.3a allows for storing or retrieving pallet loads on flow racks. Some racks are push-back and allow loading from the front. In this case, items are retrieved using the last-in, first-out policy. There are stability as well as safety concerns when handling heavy unit loads, thus, the forklifts cannot typically operate more than three levels. Note that the forklift shown in Fig. 15.3b allows the operator to stop at the required pick stations, pick the required types and quantities of items and complete one or more full orders in one trip. Figure 15.3c, d illustrate two types of automated storage and retrieval systems (AS/RS). As shown in the figures, the operator has an enclosed platform that itself is mounted on a mast and can travel up and down the mast using an electric motor. Another set of motors propels the mast in a forward and reverse direction allowing an AS/RS crane (and the operator on the platform) to access any item in an aisle. Figure 15.3e, f show two types of fully automated AS/RS. In Fig. 15.3e, the red autonomous vehicles pick up an inbound pallet from the staging area, travel on elevators as necessary to reach the desired storage level, can travel in orthogonal directions using two sets of motors, and either store the pallet or retrieve an outbound pallet using forks that extend in and out, reversing the afore-mentioned travel steps. The system shown in Fig. 15.3f has mini robots that can retrieve (or deposit) totes in storage columns, lift them up, and deposit them in another column designated for outbound totes. The materials-handling systems shown in Fig. 15.3 are just six examples of thousands of alternate devices, systems, and configurations. The reader is encouraged to visit websites such as www.mhi.org or www.cicmhe.org to learn more about materials-handling manufacturers, systems integrators, warehouse designers, and the products they offer.

15.2.3 Simple Model for Warehouse Design In this section, a basic model (Model 1) for determining the warehouse dimension (rough footprint) given the number of required storage locations, is presented. It is assumed that the unit of storage is a pallet or a tote, but that all the units are of the same size and shape. The number of storage locations depends upon the maximum

336

S. Heragu

(a) Fork-lifts used for storage and retrieval (MHI)

(b) Fork-lifts used for order picking (Crown Corporation)

(c) Top view of a person-on-board (d) Person-on-board AS/RS for pallet storage automated storage and retrieval system (Vanderlande Industries) (AS/RS)(Vanderlande Industries)

(e) Fully automated warehouse using autonomous vehicles for pallet storage and retrieval (Savoye Logistics).

(f) Fully automated warehouse using autonomous vehicles for tote storage and retrieval (Swisslog).

Fig. 15.3 Warehouse material-handling systems

inventory the warehouse manager expects the warehouse to hold for each of the items. The model can be formulated to minimize the total (or the average) one-way or roundtrip distance from one or more customer entry points located along the periphery (or the interior) of the warehouse, assuming a given location is as likely to be visited as any other. The notation used in the model in which average one-way distances are minimized from an entry point near the lower-left corner of the warehouse and

15 Warehousing

337

Fig. 15.4 Sample footprint of a members-only warehouse store

C C an exit point at the lower right-corner of the warehouse are provided below. This model (also see Fig. 15.4) can be used to determine the warehouse dimensions for a members-only warehouse store, e.g., Costco, as illustrated in Heragu (2016). Parameters N Desired number of storage locations h Height of warehouse measured in shelf height a Horizontal aisle space multiplier, measured as a multiple of the width of a storage location b Vertical aisle space multiplier, measured as a multiple of the width of a storage location Decision Variables x Number of columns of storage racks in the horizontal dimension of the warehouse y Number of rows of storage racks in the horizontal dimension of the warehouse

338

S. Heragu

Model 1 Minimize x(a + 1)/4 + y(b + 1)/2

(15.1)

Subject to x yh ≥ N

(15.2)

x, y ≥ 0 and integer

(15.3)

The model assumes that there are two entry/exit points as illustrated in Fig. 15.4 and the objective function minimizes the average, one-way distance traveled by a customer. Note that it is assumed that the customer enters from the lower-left corner of the warehouse and exits at the lower right corner, which has two locations marked with a C for two cashier posts. The horizontal aisle space multiplier a for the warehouse configuration shown in Fig. 15.4 is 2/10  0.2. Note that there are two vertical aisles and ten columns of rack spaces. Similarly, b is 6/12 = 0.5. Assuming that the constraint is an equality at optimality, i.e., xyh = N, and using this equation to define y as N/xh, it can be seen that the model can be transformed into a single variable, unconstrained model. It can be shown to be convex by observing that the second derivative of the unconstrained model is greater than 0. Thus, the optimal value of the decision variable x can be found by taking the first derivative of the objective function and setting it equal to 0. This yields the following value for x.  2N (b + 1) x h(a + 1) Substituting x = N/yh, the following equation for y is obtained.  N (a + 1) y 2h(b + 1) Assuming the warehouse has orthogonal aisles, the above model creates a footprint that minimizes the total or the average distance traveled by an order picker.

15.3 Case Study: Unilever Warehouse Beauvais, a small town about 50 miles north of Paris, France houses a large distribution center of Unilever. This facility is almost fully automated. The reserve storage in this facility (see Fig. 15.3e) has six tiers (levels) and can hold 50,000 pallets. DHL trucks arrive each morning and the truck drivers unload the inbound pallets using forklifts in a staging area (Fig. 15.5a). Then, a series of autonomous vehicle storage and retrieval systems (AVSRS) pick up the pallets and take them to a designated area (see AVSRS that has just deposited a pallet in its storage location in Fig. 15.5b). If the storage location is

15 Warehousing

339

not on the ground floor, the AVSRS must utilize the elevator shown in Fig. 15.5c to travel to the tier in which the load must be deposited. Once the designated tier has been reached, the AVSRS exits the elevator and travels on sets of rails (see Fig. 15.5b) that are orthogonal to each other. The AVSRS has two sets of motors, one to propel it in the x-direction and another in the y-direction. (The z-direction movement is provided by the elevator). It also has forks that extend in or out to retrieve or store pallets. In another section of the warehouse, a robot builds customer orders using the following approach: • It requests full pallets from the reserve section and breaks the pallet into layers. • The required number of layers of multiple stock keeping units (SKUs) are built on an empty pallet layer by layer, until it reaches the maximum height allowed (3 m) or a customer order has been filled, whichever occurs first. • The completed orders are shrink-wrapped to form a unit load. • An AVSRS picks up the unit load and places them in an outbound staging area. • A DHL truck driver then loads outbound pallets on outbound trucks (typically in the evening hours). The AVSRS are controlled by Savoye Logistics engineers remotely from Lyon, 500 km south of Paris. The control system located in Lyon allows Savoye Logistics to commission or decommission AVSRS depending upon the workload and desired throughput or for maintenance/repair purposes. Thus, using an efficient design and advanced automated materials-handling equipment, Unilever manages its warehouse operations using employees from DHL and Savoye Logistics without utilizing its own personnel, other than supervisors who visit the warehouse periodically to troubleshoot, supervise or answer questions that DHL and Savoye Logistics may have.

15.4 Warehouse Design, Storage, Routing, and Travel Policies (Advanced) In this section, a few advanced models for the design and analysis of warehousing materials handling systems are presented. Materials handling and warehousing are mature subject areas and the reader is referred to the review papers mentioned in the Further Reading section.

340

S. Heragu

(a) Fork-lift operators unloading pallets at a staging area (Savoye Logistics).

(b) AVSRS tunneling on rails (Savoye Logistics).

(c) AVSRS on an elevator returning from a higher-level tier after storing a pallet (Savoye Logistics). Fig. 15.5 Illustration of automated materials-handling systems in a warehouse

15.4.1 Advanced Model for Warehouse Design Heragu et al. (2005) present a model to determine an SKU’s allocation to areas in a warehouse. It is assumed that a warehouse has three areas—reserve storage, forward storage, and cross-docking (see Fig. 15.6). There are also receiving and shipping areas, which use a set of docks for the goods to be loaded to or unloaded from a truck. The same area and set of docks could be used for receiving and shipping (for example, receiving in the morning and shipping in the night) or there could be separate areas (and docks) for shipping and receiving. The former is typical. Full pallets that stay in the warehouse for an extended period, e.g., two weeks to two months, are typically stored in the reserve storage area. On the other hand, the forward area is

15 Warehousing

341

Fig. 15.6 Typical product flows in a warehouse (Reproduced from Heragu et al. (2005))

mainly used for case picking, accumulation, sortation, packaging, and other valueadded services performed at the warehouse. Goods are stored in the forward area for a very short period of time—hours or days. A third area in the warehouse is set aside for cross-docking. For the three areas identified in the warehouse, four flows can be identified as shown in Fig. 15.6 and detailed below. • In flow 1, pallets or cartons unloaded from an inbound truck are sorted and routed to an outbound truck. Items that are cross-docked do not spend any time on the shelves in the warehouse. • In flow 2, full pallets or cartons which are unloaded from trucks are received and stored in the reserve storage area. When a demand occurs for a full pallet load (typically after some length of time), they are shipped as full pallets. • In flow 3, pallets are received from the inbound trucks and stored in the reserve area. However, the forward area may request a full pallet load from reserve storage, use a part of it for fulfilling a given order, and then send the remainder of the pallet back to reserve storage. The number of times pallets are requested from the reserve to forward areas depends upon the demand rate for the items on the pallet and will occur until all the items on the pallet are depleted. The items that are picked at the forward area are used to put together a customer order and shipped along with other items to the customer. • In flow 4, full pallets or cartons are received from an inbound truck, stored temporarily in the reserve storage area, used to fill customer orders that include the corresponding SKU, and then shipped out as part of an order. The model presented next not only assigns individual SKUs to the three areas, but as a result, determines the size of each area and thus, provides a layout of the warehouse. It makes several reasonable assumptions and the reader is referred to Heragu et al. (2005) for additional details. Model 2 and the corresponding notation are presented next. Parameters i j

Type of SKU, i = 1, 2, …, n Type of flow, j = 1, 2, 3, 4

342

S. Heragu

Di Ai Pi Si di pi

Annual demand for SKU i Order cost for SKU i Unit load price for SKU i space occupied by a unit load of SKU i ratio of the unit load size in the reserve area to that in the forward area average percentage of time a unit load of product i spends in the reserve area when assigned to flow 3 1 if SKU i is assigned to flow j, j = 1, 2 or 4 qij qij [di ] + 1 if SKU i is assigned to flow j = 3, where di is the smallest integer greater than or equal to di a, b, c Number of vertical levels of storage in the cross-docking, reserve, and forward areas, respectively r Inventory carrying cost rate Cost of moving a unit load of SKU i in flow j H ij Annual cost of storing a unit load of SKU i in flow j C ij Time spent by SKU i in the warehouse in years Ti TS Total available warehouse floor space α, β, γ Proportion of the total space assigned to the cross-docking, reserve, and forward areas, respectively Decision Variables x ij 1 if SKU i is assigned to flow j; 0 otherwise y Number of rows of storage racks in the horizontal dimension of the warehouse Model 2 Minimi ze

n i1

4 j1

4

Subject to n

i1

(15.4)

xi j  1, i  1, 2, . . . , n

(15.5)

0.5Q i Si xi1 ≤ αaT S n 0.5Q i Si xi2 + 0.5 pi Q i Si xi3 ≤ βbT S i1 i1 n 0.5(1 − pi )Q i Si xi3 + 0.5Q i Si xi4 ≤ γ cT S

n n

j1

(2qi j Hi j Di xi j + 0.5qi j Ci j Q i xi j )

i1

i1

(15.6) (15.7) (15.8)

α+β +γ ≤1

(15.9)

xi j  0 or 1, i  1, 2, . . . , n; j  1, 2, 3, 4

(15.10)

The parameter Qi in the above model is the economic lot size and is determined using the well-known lot-size formula (see also Chap. 12), i.e.  2 Ai Di Qi  r Pi

15 Warehousing

343

The objective function (15.4) minimizes the total annual cost of moving the SKUs into and out of their respective areas as well as their storage cost. Constraint (15.5) ensures each SKU is assigned to one of the four flows. Constraints (15.6), (15.7), and (15.8) ensure that the SKUs assigned to each area are such that their collective space requirement does not exceed the allotted cubic space. Constraint (15.9) ensures the total space assigned to the three areas does not exceed the available space. Constraint (15.10) are binary variables ensuring each SKU is assigned to a flow. Note that, where unit loads in flows 1, 2 and 4 are generally pallets, they may be different in flow 3 because smaller entities (e.g., cases or boxes) are picked in the forward area, after which the (partially loaded) pallet may then be returned to the reserve area. This may be repeated several times until the pallet is depleted, causing additional handling efforts and costs. The reader is referred to Heragu et al. (2005) for additional details on and solution approaches for the model. This model is comprehensive, uses readily available information, and not only allocates SKUs to flow, but also determines the space for reserve, forward, and cross-docking areas in a warehouse.

15.4.2 Storage Policies In general, there are three types of storage policies seen in practice. They are: random, dedicated, and cube-per-index storage policies. With a random storage policy, items are stored randomly at open and available storage locations. Such a policy is appropriate for automated warehouses because the storage location and item information are stored electronically and can be readily accessed by the automated devices, which can travel to the corresponding pick location without wasting any time looking for that item. In a manual system, the operator would have to search for that item, hence the dedicated storage policy is preferred. In this policy, a set of storage locations is designated for each SKU, based on its turnover. With a dedicated policy, each location can only contain the item designated to be stored in that space. Thus, although an arriving item may not have an available space in its designated location, it cannot be stored in another SKU’s location, even if that location is open and available. The dedicated policy, therefore, requires more storage spaces for the same types and volumes of SKU handled. The cube-per-order index (COI) policy is between the two policies. While each item has its set of designated spaces, these spaces are located near the warehouse entry/exit points based on their size and frequency of usage—hence the name cube (size) per order (frequency). This policy essentially calculates the ratio of each SKU’s size to the frequency with which it is demanded (e.g., annual demand) and ranks the SKU’s in non-decreasing order of their COIs. The first SKU in the list is then assigned to the set of closest spaces required by that SKU, then the next SKU in the rankordered list is assigned to the next set of closest spaces required by that SKU, and so on, until all the SKUs are assigned. Next, a mathematical model (Model 3) for

344

S. Heragu

the COI storage policy is presented. The model determines the spaces to which each SKU must be assigned. It uses the following notation. Parameters i j fi Si wj

Type of SKU, i = 1, 2, …, m Storage location, j = 1, 2, …, n frequency of usage of SKU i number of storage locations required for SKU i distance of storage location i to entry/exit point of the warehouse

Decision Variables x ij 1 if SKU i is assigned to storage location j; 0 otherwise Model 3 m n f i w j xi j Minimi ze i1 j1 Si n Subject to xi j  Si , i  1, 2, . . . , m j1 m xi j  1, j  1, 2, . . . , n i1

xi j  0 or 1, i  1, 2, . . . , m; j  1, 2, . . . , n

(15.11) (15.12) (15.13) (15.14)

The objective function (15.11) minimizes the total distance of moving the SKUs into and out of their respective areas. Constraint (15.12) ensures each SKU is assigned the required number of storage locations. Constraint (15.13) ensures that each storage location is assigned only one SKU. Constraint (15.14) is a binary constraint. The reader is referred to Heragu (2016) or Malmborg and Bhaskaran (1990) for additional details on the model. If a random storage policy is used to store goods in a warehouse, for example, one in which the SKUs are identical or their sizes are uniform (e.g., pallet loads) and the SKU locations are readily determined via electronic means (e.g., pick-tolight systems), then an efficient warehouse layout is one which minimizes the travel distance for the order-picker. If it is assumed that each location is equally likely to be visited in an order picker’s tour, then depending upon the number of entry/exit points, storage locations, and the travel distance metric, the warehouse shape can be anything from rectangular to triangular to trapezium to a circle. For example, if it is assumed that there is only one entry/exit point in the middle of the south wall in the warehouse shown in Fig. 15.7, which must have 30 storage locations, and that each storage location is 1 m by 1 m2 , then based on the rectilinear distances to the entry/exit point, the layout that optimizes travel distances is a triangular layout one that is shown by grey cells with the distance of that cell to the entry/exit point. The layout in Fig. 15.7 was determined assuming the travel distance metric was rectilinear. In other words, the aisles are assumed to be orthogonal. Gue and Meller (2009) essentially took the same idea and flipped the question. If the warehouse shape is fixed, what should the aisle structure be assuming the travel distance metric

15 Warehousing

10

10 10

345

10 10 10

10 10 10 10

10 10 10 10 10

10 10 10 10 10

10 10 10 10

10 10 10

10 10

10

Fig. 15.7 Warehouse layout assuming random storage, entry/exit point in the middle of the South wall, and 30 storage locations

Fig. 15.8 Flying V aisle design (adapted from Gue and Meller (2009))

is rectilinear? Based on the assumptions, they propose cross-diagonal aisles that are not necessarily orthogonal. In fact, they are at 45° angles to the exterior walls of the warehouse. They show that under certain conditions, alternate versions of the aisle structure that include piecewise, diagonal cross-aisles, reduce travel by as much as 20%. An example of the aisle structure is shown in Fig. 15.8. The reader is referred to Gue and Meler (2009) for additional details.

15.4.3 Routing Strategies The routing policy adopted by order pickers in a warehouse can have an impact on their efficiency. Specifically, the routes undertaken by the order pickers in a warehouse can contribute to their throughput. In a warehouse with vertical (or horizontal) aisles and orthogonal cross-aisles, there are many ways in which an order picker can traverse the aisles. For example, consider the warehouse layout shown in Fig. 15.9a. The highlighted cells represent the items that must be picked in a customer order. Multiple routes can be taken to pick all the items in an order. One could travel aisle

346

S. Heragu

(a) Warehouse with seven items to be picked.

(b) S-Shaped routing method.

(c) Largest gap method.

(d) User-generated routing.

(e) Optimal method. Fig. 15.9 Illustration of some of the routing methods available for order-picking (generated using the ‘Interactive Warehouse’ tool available at roodbergen.com)

15 Warehousing

347

Fig. 15.10 Summary of results for various routing methods for the example in Fig. 15.9

by aisle, picking all the items in an aisle from closest to farthest. This is called the aisle-by-aisle method. Many others are possible, including the S-shape, largest gap, and others. The S-shaped method requires the order picker to traverse an aisle if there is at least one item to be picked. The order picker goes to the farthest item in the aisle, makes a right angle turn and proceeds until he or she reaches an aisle, which contains at least one item to be picked. Once again, the order picker makes a right angle turn and proceeds to the next aisle and repeats the afore-mentioned traversing procedure until all the items are picked and then returns to the order collation station. There are some nuances with this method that are explained in roodbergen.com where there is also an interactive tool that allows users to experiment with alternate routing strategies, including ones they can create for specific warehouse configurations and randomly generated or user-generated pick locations in an order. The S-shaped method for the order list in Fig. 15.9a is shown in Fig. 15.9b. Loosely explained, the largest gap heuristic does not necessarily traverse an entire aisle if indeed, a closer item in the order picker’s list exists in an adjacent aisle. This process is repeated until all the items in an order are picked (see Fig. 15.9c). A usergenerated routing is shown in Fig. 15.9d. Lastly, the optimal algorithm described in Roodbergen and De Koster (2001) is illustrated in Fig. 15.9e. It determines the shortest path based on the algorithm in Ratliff and Rosenthal (1983). Details are available at roodbergen.com. A summary of the results obtained using the methods described in Fig. 15.9 is provided in Fig. 15.10. Note that of the two user-generated routes, one is better than the largest gap method, but neither is as good as the optimal route.

348

S. Heragu

15.5 Current Topics that May Shape the Future of Warehousing (State of the Art) 15.5.1 Trends Impacting Warehouse Design, Function, and Operations There are many trends occurring that will change the size, function, character, and nature of warehousing. Some of these are listed below. • As mentioned in the Introduction section, companies such as Ali Baba do not own a warehouse, yet are one of the largest (if not the largest) retailers. They do this by serving as a virtual link between the product manufacturers and customers. • Mass customization and direct orders by customers on online platforms will mean that quantities will be shipped in small quantities, perhaps in lot size of 1! • The share of brick-and-mortar stores is constantly declining ceding more business to online stores. • Environmental concerns may force companies to build fewer, but large warehouses on small footprints. Thus, warehouses that are 50 m or more in height could become commonplace and more automated materials handling equipment that can handle unit lot sizes could become important. • Driverless and unmanned aerial vehicles will play a significant role in the last-mile delivery. • There will be many research problems that will need to be addressed in the area of last-mile delivery. • The desire to source from local suppliers, for all products, but especially for agricultural products, will mean that shipping distances will decrease and there will be a reduced need for container ships. • Virtual reality (VR) will reduce the need for the manufacture, storage, and distribution of some physical products because a consumer will be able to get the same experience with VR, eliminating the need for the corresponding physical product. • Sensors are ubiquitous and the ability for wirelessly connected devices to communicate will have an impact on how, when, and where goods are stored. For example, the data transmitted from sensors on refrigerators may dictate how replacement parts are ordered and stocked in a nearby warehouse.

15.5.2 Materials Handling Roadmap The materials handling industry, through MHI, organized four workshops in Atlanta, Chicago, Los Angeles, and Washington D.C., involving experts in the field of material handling and logistics from academia, industry and government. This set of meetings and additional work produced a material handling and logistics roadmap that is available at http://www.mhlroadmap.org. The workshop participants were tasked

15 Warehousing

349

with forecasting what the industry would look like in the year 2025 and to identify supporting trends, challenges, and capabilities faced by the material handling and logistics industry (Gue et al. 2014). The reader is encouraged to review the roadmap to understand the various trends, challenges, and capabilities that are likely to be faced the material handling and logistics industry.

15.5.3 Warehouse Design Conceptualization Simulation is the preferred modeling tool for warehouses designers to develop configurations, make modifications, and develop final designs that meet the customer’s requirements. Simulation is also used to demonstrate that the chosen final configuration does indeed meet the throughput requirements. However, a disadvantage of this approach is that it takes a significant amount of effort and time to develop, test, verify, validate, and run a simulation model. As a result, only a handful number (one, two or three) of alternate configurations are considered. Each is tested and the best of the two or three designs is then presented as a solution to the customer. Thus, for example, layouts with only two to three alternate footprints may be considered in designing the warehouse mentioned in Sect. 15.3. It could be that hundreds of footprints may need to be explored before settling on a final layout. This cannot be done with simulation because of the effort and time it takes to build and test one model. Queuing network models on the other hand can be used to provide approximate estimates of key performance measures of hundreds of alternate design configurations very quickly. A designer could then examine each, see which of these designs has the ability to yield the best performance, select that design and develop a detailed simulation model for that design. This approach will yield a much better final result than the one currently adopted by practitioners. Heragu et al. (2009) present open queuing network models as a tool to determine whether the AS/RS or AVSRS is preferred for a given design scenario. Cai et al. (2014) model alternate versions of AVSRS and develop a semi-open queuing network model to determine the preferred version to use. Other models such as those in Roy et al. (2015) and those in the Further Readings section are recommended for readers who want to learn more about these approaches.

15.5.4 Further Reading The literature is rich with papers on the design and analysis of warehousing systems and some of the more important problems in each category, e.g., AS/RS and order picking. In addition to the afore-mentioned articles, a number of survey articles exist in the literature. The reader is recommended to review the papers in the References section so he or she can gain an understanding of other topics and papers not covered in this chapter.

350

S. Heragu

References Askin RA, Standridge CR (1993) Modeling and analysis of manufacturing systems. Wiley, New York, NY Cai X, Heragu SS, Liu Y (2014) Modeling and evaluating the AVS/RS with tier-to-tier vehicles using a semi-open queueing network. IIE Trans 46(9):905–927 Gue KR, Meller RD (2009) Aisle configurations for unit-load warehouses. IIE Trans 41(3):171–182 Gue KR, Akcali E, Erera A, Ferrell W, Forger G (2014) U.S. Roadmap for Materials Handling and Logistics. Materials Handling Institute, Charlotte, NC Heragu SS (2016) Facilities design. CRC Press, Boca Raton, FL Heragu SS, Cai X, Krishnamurthy K, Malmborg CJ (2009) Analytical models for analysis of automated warehouse material handling systems. Int J Prod Res 49(22):6833–6861 Heragu SS, Du L, Mantel RJ, Schuur PC (2005) Mathematical model for warehouse design and product allocation. Int J Prod Res 43(2):327–338 Hu W, Li Y, Govindan K (2014) The impact of consumer returns policies on consignment contracts with inventory control. Eur J Oper Res 233(2):398–407 Malmborg CJ, Bhaskaran L (1990) A revised proof of optimality for the cube-per-order index rule for stored item location. Appl Math Model 14(2):87–95 Ratliff HD, Rosenthal AS (1983) Order-picking in a rectangular warehouse: a solvable case of the traveling salesman problem. Oper Res 31(3):507–521 Roodbergen KJ, De Koster R (2001) Routing methods for warehouses with multiple cross aisles. Int J Prod Res 39(9):1865–1883 Roy D, Krishnamurthy A, Heragu SS, Malmborg CJ (2015) Queuing models to analyze Dwellpoint and cross-aisle location in autonomous vehicle-based warehouse systems. Eur J Oper Res 242(1):72–87 Tompkins JA, White JA, Bozer YA, Frazelle EH, Tanchoco JMA, Trevino J (2010): Facilities planning, 4th edn. Wiley, New York

Further Reading Gu J, Goetschalckx M, McGinnis LF (2007) Research on warehouse operation: a comprehensive review. Eur J Oper Res 177:1–21 De Koster R, Le-Duc T, Roodbergen KJ (2007) Design and control of warehouse order picking: a literature review. Eur J Oper Res 182:481–501 Manzini R, Bozer YA, Heragu SS (2015) Decision models for the design, optimization and management of warehousing and material handling systems. Int J Prod Econ 170:711–716 Roodbergen KJ, Vis IFA (2009) A survey of literature on automated storage and retrieval systems. Eur J Oper Res 194:343–362 Rouwenhorst B, Reuter B, Stockrahm V, van Houtum GJ, Mantel RJ, Zijm WHM (2000) Warehouse design and control: framework and literature review. Eur J Oper Res 122:515–533 Roy D, Krishnamurthy A, Heragu SS, Malmborg CJ (2014) Blocking effects in warehouse systems with autonomous vehicles. IEEE Trans Autom Sci Eng 11(2):439–452 Roy D, Krishnamurthy A, Heragu SS (2015) Stochastic models for unit-load operations in warehouse systems with autonomous vehicles. Ann Oper Res 231(1):129–155 Roy D, Krishnamurthy A, Heragu SS, Malmborg CJ (2016) A Simulation framework for studying blocking effects in warehouse systems with autonomous vehicles. European J Indust Eng 10(1):51–80 Roy D, Krishnamurthy A, Heragu SS, and Malmborg CJ (to appear in 2017) A multi-tier linking approach to analyze performance of autonomous vehicle-based storage and retrieval systems. Comput Oper Res

15 Warehousing

351

Sarker BR, Sobhan Babu P (1995) Travel time models in automated storage and retrieval systems: a critical review. Int J Prod Econ 40:173–184 Tompkins JA, White JA, Bozer YA, Tanchoco JMA (2010) Facilities Planning, Wiley, New York, NY

Chapter 16

Closed Loop Supply Chain Management Erwin A. van der Laan

Abstract This chapter provides an overview of concepts within the field of closed loop supply chain management. First, using some qualitative frameworks we categorize and characterize the many different closed loop supply chains that are found in practice. Then, we identify key processes and potential bottlenecks that need to be addressed for efficient and effective management. Several modeling approaches are presented, moving from very basic via more advanced to state-of-the-art models that make complex trade-offs and generate more refined insights. The mathematical models presented are meant as simple illustrations of broader concepts. Hence, this chapter is certainly not meant as a comprehensive review of the logistics and operations management literature within the field. Rather, we hope that it can inspire newcomers to the field and specialists who would like to broaden their view.

16.1 Fundamentals of Closed Loop Supply Chain Management (Basic) Our current society more and more recognizes the limits of our linear economy: an economy that basically transforms inputs into waste. Besides the related environmental and social issues, it also seems a missed opportunity in terms of business value creation. Its counterpart, the circular economy, is based on the principle that waste should not exist, as after use products and materials should somehow be inputs to a new cycle of production and use. Hence, on the micro-economic level the challenge is to turn linear supply chains into circular ones: closed loop supply chains (CLSCs). Circular supply chains are not a new phenomenon. Actually, only because of the industrial revolution with its double exponential growth in technology and welfare does the developed world have the ‘luxury’ to waste what it produces. Currently, on

E. A. van der Laan (B) Rotterdam School of Management, Erasmus University, Rotterdam, The Netherlands e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_16

353

354

E. A. van der Laan

the material level, we have gotten used to separate waste paper and glass collection and recycling. On the product level there are second-hand shops and online trading and selling platforms, such as Ebay and Amazon. However, these examples comprise only a very small portion of total worldwide production. To give a couple of examples, of the total usage of copper, only some 35% comes from recycled copper.1 Worldwide we generate 42 million tons of electronic waste of which only about 6.5 tons is recovered by official take back systems2 ; a substantial portion of this waste ends up in third world countries where it poses a serious threat to the environment in general and human health in particular. Although recycling decreases waste, it is a rather low level of recovery that only recycles the materials, but not all the other inputs that went into manufacturing the original product, such as energy, water and labor. For instance, the manufacturing of a single car requires some 150.000 L of clean water3 and an amount of energy that is worth about three years of driving.4 These inputs are not recovered by car recycling. Hence, from both an environmental and economical point of view it may make more sense to recover on a product level (for instance, refurbishment of cell phones) or component level (for instance, refurbishment of car engines). The latter two examples are already commercially available on a relatively large scale; for most products though, there is no scale to speak of.

16.1.1 A Framework for Closed Loop Supply Chains It is rather difficult to speak of closed loop supply chains in general as they come in so many forms. To appreciate this, let’s consider the framework of de Brito and Dekker (2004) that characterizes a closed loop by answering the following five questions. 1. What is the nature of the product that is returned? The composition and complexity of what is returned determine how feasible it is to close the loop. Products that consist of a single material may be very easy to recycle. Products that are a complex assembly of many components containing many materials may be very hard to recover on the product and material level. 2. What is the reason that it becomes available for recovery? Whether it is a commercial return that was never used, a warranty return that may be faulty, an end-of use-return that is still in good condition, or an end-of-life return that is no longer working, determines the recovery options that are available. In general, the better the quality, the more options are available and the more value can be retrieved in the recovery process. 1 World

Copper Factbook 2013, International Copper Study Group, p. 50. global E-waste monitor 2014, United Nations University, pp. 22–23. 3 According to the United States Environmental Protection Agency (EPA) it takes 39,090 gallons (EPA report 810-F-95-001, 1995). 4 M. Berners-Lee and D. Clark in The guardian “What’s the carbon footprint of ... a new car?”, 23 September 2010. 2 The

16 Closed Loop Supply Chain Management

355

Fig. 16.1 Key processes of the CLSC

3. What is the reason that a supply chain actor actually recovers the product? The main driver for recovery, be it economics, legislation or corporate social responsibility, very much determines the type of recovery network. Legislation drives efficient supply chains for recycling, where input quality and time play a minor role; for economically driven networks it can be quite the opposite. 4. How does the CLSC work in terms of process? The key processes of the CLSC are, acquisition, recovery and remarketing, connected by the end customer, who is both the consumer and supplier of products. Products that are not effectively collected and/or recovered may end up as waste (Fig. 16.1). Each of these processes can be organized in different ways with possibly different consequences for effectiveness and efficiency. Each may come with various economical, ecological or technological challenges that may form bottlenecks in the closed loop. As these three processes are paramount for the design of the closed loop supply chain, we take these processes as an additional structure to discuss the basic, advanced and state-of-the-art concepts of CLSC modeling. 5. Who is involved in closing the loop? The CLSC is more complex than a traditional supply chain as it typically involves additional stakeholders, such as specialized collectors, recyclers and resellers. Many interactions may exist between the forward chain (new production and forward distribution) and the reverse chain (recovery and collection). The party that takes the lead determines to a large extent the nature of the CLSC. If it is the original manufacturer, we often see a so called closed closed loop (that is, the product returns to its origin). If the initiator is a third party that is not part of the original forward chain, we usually find an open closed loop (that is, the recovered product enters a new supply chain). Recycling chains are typically open closed loops.

356

E. A. van der Laan

Clearly, the context (nature and reason of the product return) plus the driver to close the loop determine to a great extent what the configuration of the CLSC should be. Next, we classify these possible configurations.

16.1.2 CLSC Configurations Fleischmann (2003) identifies five different configurations of reverse logistics networks to close the loop, depending on the supply chain driver (economics, legislation), recovery method (recycling, value added, direct reuse) and instigator of the closed loop (brand owner, opportunistic player). This framework, however, does not cover the important category of commercial and warranty returns that are initially handled by the retailers. Adding the latter we have the following six categories: 1. Legislation driven recycling network: Collection and recovery is mandated by a legislative framework. Collection and recovery activities are usually executed by parties that are not affiliated with the originating chain, but installed and monitored by governmental institutions. Examples are networks for electronic waste recycling, end-of-life vehicle recycling and battery recycling as mandated by directives of the European Union. 2. Profit driven recycling network: If material recycling is profitable, original supply chain actors or opportunistic players may initiate collection and recycling activities. Typically these are specialized recyclers as investments in recovery facilities are high. 3. Profit driven value added network by opportunistic player: If value added recovery is profitable, opportunistic players may initiate collection and recycling activities independently of the original manufacturer. Examples are independent cell phone recovery networks and independent ink and toner cartridge recovery networks. 4. Profit driven value added network by brand owner: Sometimes brand owners themselves initiate the value added recovery. In principle, the brand owner has the advantage over opportunistic players as it has access to the original product design and it can use its existing distribution infrastructure and manufacturing facilities to do the collection and recovery. Examples are spare part remanufacturing networks and networks for refurbished consumer electronics (laptops, tablets, etc.). 5. Direct reuse network for distribution items: This is a special category as it facilitates the logistics of products. The items (containers, pallets, etc.) are valuable, so worthwhile to collect, reuse and repair if necessary. The ownership and management may reside at pool operators. 6. Direct reuse network for commercial and warranty returns. Enforced by consumer rights, retailers have the legal obligation to take back sold products and refund the customer within a certain time window (commercial returns). The

16 Closed Loop Supply Chain Management

357

same holds for products that the consumer claims to be faulty (warranty returns). A large portion of these product return flow is of very high quality or as good as new, so that they can be refurbished or restocked. Careful and timely collection and recovery can retain much business value that would otherwise be lost. Even when closed loops are driven by legislation or corporate responsibility, these initiatives will only be feasible at sufficient scale in the long run if the underlying CLSC is economically feasible. Hence, in the next section we discuss the different types of business value that can be generated by closing the loop.

16.1.3 CLSC Business Value To assess the profitability of a CLSC, there is more to it than just looking at the direct financial impacts. In fact, the following types of business value that can be generated by closing the loop must first be identified (Schenkel et al. 2015). • Sourcing value: These are the direct financial impacts, such as cheaper sourcing of products, components and materials and avoidance of negative externalities, such as disposal fees and environmental fines. • Environmental/social value: Indirectly, communicating product recovery initiatives to relevant stakeholders may lead to improved brand image and easier compliance with (future) legislation. • Customer value: Through product recovery activities, a wider array of products and services can be offered, which may lead to higher customer satisfaction and loyalty. • Information value: Actively collecting and inspecting product returns can provide essential information to improve product design and optimize forward and reverse logistics processes. Hence, when closing the loop, one should holistically consider all four value types in order to decide on the best strategy. Given the best strategy, one can begin to think about optimizing the closed loop supply chain on the tactical and operational levels.

16.2 Key Concepts in Closed Loop Supply Chains (Basic) In this section we focus solely on the reverse flows of the closed loop chain without any interaction with the forward flows. This is typically the case when opportunistic players not associated with the originating chain close the loop. The case where there is interaction is discussed in Sect. 16.3.

358

E. A. van der Laan

Case 1: Cell Phone Recovery by Gazelle Gazelle is one of the most popular trade-in platforms for used consumer electronics like old cell phones, tablets and computers in the US. Consumers can offer their used or broken electronic devices in return for cash. Online, consumers enter brand, model and some simple indicators for device quality (working or not, scratched or not) and immediately receive a sell back price quote. The device is shipped for free and if after inspection the device is indeed worth the quoted price, the money is sent via PayPal or gift cards at Amazon.com. Even faster is to return the device at one of the automated ecoATM kiosks. The phone is entered into the kiosk and after an automated external inspection, the phone is plugged in so that an internal inspection can be performed, of course without accessing personal information. After that, the consumer receives a quote. If the customer accepts the quote, the phone can be handed to the kiosk and the transaction is completed. The company reached $100 million in revenue in 2013 with a growth rate of 150% (Kirsner 2014). By 2014 Gazelle had bought back 2 million devices from 1 million consumers for a total amount of $200 million averting some 50.000 tons of e-waste from landfill (Lyons Hardcastle 2014). Recovery of cell phones is relatively simple. If the phone works technically, then only some inexpensive parts need to be replaced, such as battery, casing and screen. The most important process is data wiping to guarantee the privacy of the previous user and avoid undesired surprises for the next user. Phones that cannot be refurbished are recycled at certified recyclers, although this does not contribute to Gazelle’s profits. There is ample demand for refurbished phones, particularly in developing countries. There, phone network coverage is usually not the problem, but people do not have the financial means to buy the latest newly manufactured cell phones. The main challenge is to acquire a sufficient volume of used phones. To accomplish this, companies like Gazelle make use of a vast network of suppliers (cell phone carriers, OEMs, retailers, online trade-in, enterprises, charity organizations). Although the heterogeneity of the product return flow is high (the variety in cell phones at any time easily reaches 600 models), the number of spare parts to put on stock is very limited per model and parts are small in size.

16.2.1 Acquisition The closed loop supply chain stands or falls with the availability of reusable products in the right quality and quantity at the right time for a reasonable cost. Usually, an

16 Closed Loop Supply Chain Management

359

active acquisition process is necessary to counter the uncertainty that is typically associated with used products. In particular when the closed loop is initiated by a third party that is not affiliated with the original manufacturing chain, suppliers of reusable products can be many and vary considerably in terms of offered quantity, quality, price and transaction conditions (e.g., the Gazelle case). In order to guarantee sufficient output of the recovery process, one needs to be able to forecast the availability of reusable products, sort them into quality categories and distribute them to the right disposition option (product recovery, part recovery, material recycling).

16.2.1.1

Modeling the Flow of Reusable Products

A fundamental starting point for modeling the flow of reusable products is to make assumptions about its distribution. In reality, the product return flow is a result of the original manufacturing, distribution and usage process, but this may be too complex to model or historical data is simply not available to estimate its parameters. A simple approach is to assume that the product flow is an independent process that is completely detached from the originating supply chain and posits a reasonable distribution of the number of products per time unit. In deterministic modeling, the rate is often taken to be constant (e.g., Schrady 1967; Richter 1996; Teunter 2001; Minner and Kleber 2001). In stochastic modeling, a popular choice is the Poisson distribution (e.g., Muckstadt and Isaac 1981; van der Laan et al. 1999; Fleischmann and Kuik 2002), as it is positive valued and it allows for Markov chain modeling. Van der Laan et al. (1999) also explored the Coxian distribution to model the return flow as it allows for more flexibility in setting variance and skewness. The normal distribution is an alternative with the advantage that it is a common assumption in inventory management modeling. A drawback is that it takes on negative values as well. Often net demand (demand minus returns) per time unit is modeled, where the normal distribution is a more reasonable choice (for applications see de Brito and van der Laan 2009; van der Laan and Teunter 2006). Some approaches do not make any assumptions on the distribution (e.g., Simpson 1978; Inderfurth 1997).

16.2.1.2

Collection Inspection Sorting and Disposition

Figure 16.2 depicts a generic collection and disposition network. The level of centralization of the different activities strongly depends on the setup cost of activity centers and potential economies of scale. Depending on the type of network, some activities are more important than others. For instance, for recycling networks sorting and grading may not be necessary, while for refurbishment networks it may be essential. Specialized inspection centers may be bypassed for some products if visual inspection at the collection sites can identify products to be sent directly to

360

E. A. van der Laan

Fig. 16.2 Generic network for collection, sorting, recovery and redistribution of used products

product level recovery (very high quality) or recycling (very poor quality). This is particularly important when market prices for recovered products drop quickly in time (Blackburn et al. 2004). To avoid unnecessary reverse logistics costs, one may even want to avoid product returns in the first place (gatekeeping). Consider the available return flow Ac from collection site c, the demand for recovered products Dr at recovery center r ∈ {0, 2, . . . , R},5 the capacity C Ii of inspection center i ∈ {1, 2, . . . , I }, the capacity C Rr of recovery center r (recycling capacity C R0 = ∞), the recovery yield y, the unit collection cost kci from collection site c to inspection center i, the unit transportation cost tir from collection center i to recovery center r , the unit recovery profit sr in recovery center r and the fixed per time unit installation cost Fi of inspection center i (e.g., total investment divided by useful life of asset). Then, a basic, deterministic, single period model for optimizing the product flows X ci from collection center c to inspection center i, flows Yir from inspection center i to recovery center r and location decisions Z i regarding inspection centers i, is given as follows.

5r

= 0 indicates ‘recycling’ for which demand is assumed infinite.

16 Closed Loop Supply Chain Management Minimize

 

X ci kci +

c∈C i∈I

  i∈I r ∈R

Yir (tir − sr ) +

361 

Z i Fi

i∈I

subject to the constraints  i 

X ci = Ac

for all c (all available product returns are collected)

Yir ≤ Dr

for all r (do not recover more than demanded from recovery center r) for all i (do not exceed capacity of inspection center i)

i

 c  i 

X ci ≤ Z i C Ii Yir ≤ C Rr  Yir = y X ci

r >0

c∈C

Yi0 = (1 − y) X ci , Yir ≥ 0 Z i ∈ {0, 1}



c∈C

X ci

for all r (do not exceed capacity of recovery center r ) for all i ∈ I (only products with sufficient quality are refurbished, ...) for all i ∈ I (...the remainder is recycled) for all c ∈ C, i ∈ I , r ∈ R (non-negativity) 1 if inspection center is selected, 0 otherwise.

For the disposition decision it is essential to assess product quality. Product return quality has several dimensions (Çorbacıo˘glu and van der Laan 2013). ‘Positive intrinsic value’ relates to the inherent quality of the return (can it be recovered for a reasonable profit?). ‘Negative external value’ relates to potential damage to brand image or environmental fines that may occur if reusable products are not properly recovered. The time-based value relates to the drop in value as time passes, because market prices decrease and some recovery options become infeasible. Positive intrinsic value calls for quality differentiation to maximize total value and high time sensitivity makes this more urgent. Most modeling efforts only consider positive intrinsic value and treat the quality (distribution) as given to serve as input for the model. For instance, Das and Chowdhury (2012) consider a model for quality-based recovered product mix planning, where the quality of collected products fall in three categories (end-of-life, faulty, unused). Various papers (Aras et al. 2004, 2006; Tagaras and Zikopoulos 2008; Loomba and Nakashima 2012) consider the used product acquisition of multiple quality classes that serve as input for the remanufacturing process. The two classes, though, have different operational parameters, such as remanufacturing cost and recovery lead time, which affect decisions pertaining to the timing and quality of the purchased items. They show that taking the product quality explicitly into account may lead to significant operational savings. Tagaras and Zikopoulos (2008) show that when the quality assessment is subject to error, optimization becomes complex.

16.2.2 Disassembly and Recovery When products are recovered on the part level, product disassembly is obviously a necessary step in the recovery process. But this often also holds for recovery at the

362

E. A. van der Laan

product level, as some part may need to be replaced or repaired. When products need to be brought back to ‘as good as new’ condition, disassembly and recovery can be quite elaborate processes. Disassembly is not necessarily the reverse of assembly as the following list (adaped from Inderfurth and Teunter 2003) shows. • Uncertainty exists in terms of the configuration of the product and the quality of the subcomponents. This may even lead to stochastic routings through the disassembly/recovery shop. • Products may or may not be disassembled completely. • The disassembly process can be non-destructive or destructive, if one is only interested in harvesting some particular parts. • There may be many demand sources (potentially one for each sub-assembly), which complicates decision making. To control the disassembly process, Inderfurth and Teunter (2003) suggest five steps: 1. Identify all feasible disassembly strategies; 2. For each strategy, identify all feasible recovery options; 3. For each disassembly/recovery combination, determine the net profit; 4. Based on the net profits, determine the best disassembly/recovery strategy; 5. For all product types, schedule the disassembly/recovery activities given the demand forecasts for all products and subassemblies. Because the complexity of the above scheme is substantial, only some steps are optimized simultaneously and simplifying assumptions (e.g., deterministic demand) are made to keep the problem mathematically tractable. The most basic example of product recovery scheduling problems under the (very) limiting assumptions of one product and deterministic returns and demands is the so-called reverse Wagner/Whitin model (Richter and Sombrutzki 2000). Consider for each period t demand Dt , returns dt , end of period inventory of returned products yt , end of period inventory of recovered products It , setup cost rt , cost of holding returned product inventory h t , cost of holding recovered product inventory Ht . Then we may optimize the number of products recovered xt in each period t with the following model. Minimize

T 

rt δ(xt ) + h t yt +Ht It

t=1

subject to the constraints y0 = I0 = 0 yt = yt−1 − xt + dt It = It−1 + xt − Dt t 

(d j − D j ) ≥ 0

j=1

It , xt , yt ≥ 0

for t = 1, . . . , T (balance equation for used product inventory) for t = 1, . . . , T (balance equation for recovered inventory) for t = 1, . . . , T (ensure that in each period there are enough resources to satisfy demand) for t = 1, . . . , T (non-negativity)

where δ(x) is the indicator function (1 if x > 0, 0 elsewhere).

16 Closed Loop Supply Chain Management

363

16.2.3 Remarketing Offering recovered products to the market comes with a number of unique challenges when compared to the marketing of new products, because recovered products typically have the following unique features (adapted from Atasu et al. 2008). • • • •

The recovered product is a low-cost alternative to the original product. The recovered product is valued lower than the original product. For some consumers, the recovered product has a green image. As the recovered product has similar functionality as new products, it is often believed that it cannibalizes new product sales. • The sales of recovered products is bounded by the previous sales of the original products.

Ferrer and Swaminathan (2006) find that for the brand owner that also offers recovered products, the value of making new products in earlier time periods increases (as it creates future availability of used products), while the value of making new products in later periods decreases (as it is partially cannibalized by sales of recovered products), provided of course that recovery is profitable. Assuming there is no competition, Atasu et al. (2008) identify the conditions under which offering recovered products is profitable depending on the net unit profit of recovery, the segment size of ‘green’ consumers, market growth rate, and recovered product valuation. Ovchinnikov (2011) shows that the fraction of consumers who are willing to switch from the new to the refurbished product has an inverted U-shape. As a consequence, if the price of the recovered product is too high or too low, very few switch from new to recovered, or vice versa. Agrawal et al. (2015b) show that offering recovered products may influence the valuation of new products, which subsequently influences the brand owner’s decision to offer the recovered variant. Ferguson and Toktay (2006) allow a third-party remanufacturer to enter the market if the brand owner chooses to not offer recovered products. They show that under certain conditions it is optimal to collect and recover all available used products or none.

16.3 Interaction of Forward and Reverse Flows (Advanced) In the previous section, we limited the discussion on the reverse flows of the closed loop chain without any interaction with the forward flows. In this section we focus precisely on this interaction, which usually complicates modeling and analysis. Interaction between forward and reverse processes occurs typically when a brand owner initiates the closed loop, as the following case illustrates.

364

E. A. van der Laan

Case 2: Car Part Remanufacturing by Volkswagen (adapted from Dekker and van der Laan 2003) Volkswagen has remanufactured car parts since 1947. Starting with the beetle engine, it now remanufactures about 10.000 different car parts and refers to them as “genuine exchange parts”. Remanufacturing is a very particular recovery option in the sense that the part is brought back to as-good-as-new condition. In functionality, it is indistinguishable from a newly manufactured part and it is typically sold with similar warranty conditions. To bring it back to this as-good-as-new condition Volkswagen applies a series of extensive recovery processes such as disassembly, cleaning, repair and replacement. The recovery of engines, for instance, is done with cutting-edge technology. Car parts that fail are collected by Volkswagen via their network of car dealers. The parts are shipped to Volkswagen Germany and initially stocked. When needed, parts are sent to certified remanufacturing companies. Not all parts are recoverable, but the third party remanufacturers have a contractual obligation to deliver the same quantity of remanufactured parts. Hence, the remanufacturing parties also have their sources of failed parts that they need to use to make up for the unrecoverable portion of parts. Since both demand and supply of remanufactured parts vary and are uncertain, there may be a temporary shortage of remanufactured parts. In that case, newly manufactured parts can be used as emergency supply. This is an expensive option though, as remanufactured parts typically cost half as much as the newly manufactured parts, while margins are higher. Proper inventory management is therefore crucial, although trade-offs are rather complex due to the many processes and stocking points that are involved.

16.3.1 Forecasting the Flow of Reusable Products In order to make proper tactical and operational decisions, such as (re-)distribution capacities, recovery and manufacturing capacities, order quantities and timing, and pricing it is important to be able to forecast future return flows, especially when there is interaction between production and recovery and/or interaction between forward and return flows. Motivated by the case of container returns at Coca-Cola and Fanta, Goh and Varaprasad (1986) develop a methodology to estimate product life-cycle characteristics, such as time to return, return loss, and expected useful life. For instance, if xt and yt are the aggregated number of products sold and collected in period t respectively, then n  xt−i pi (16.1) yt = i=1

16 Closed Loop Supply Chain Management

365

with pi the probability that a product sold returns after i periods and n the maximum return window. Products that have not returned within n periods are assumed to be lost. If xt and yt are recorded and pi can be estimated, we would be able to compute important life-cycle parameters like P=

n 

pi ,

i=1

which is the fraction of sold products that eventually returns, or the average return time of a product that eventually returns n D=

i=1 (i

+ 1) pi , P

or the number of cycles that a product on average goes through before it is lost (‘trippage’) P . T = 1− P To avoid issues with multi-collinearity, Goh and Varaprasad (1986) do not apply standard linear regression to estimate the pi ’s from relation (16.1), but employ a linear function transfer instead. They recommend using monthly data for stability and at least 50 datapoints, which for monthly data would account for about four years of data. Practical complications arise, of course, if such data is not available, or if characteristics are expected to change during such a long time span. Alternative approaches are offered by Marx-Gomez et al. (2002), who suggest a combination of simulation and neural network learning, and Matsumoto and Komatsu (2015) who apply standard ARIMA and Holt-Winter models to the product return data. The latter paper reports forecasting errors of 18–27% for a dataset consisting of 160 types of car parts, showing that there still is ample room for improvement.

16.3.2 Forward and Reverse Network Design The network design problem introduced in Sect. 16.2.1 becomes a bit more complicated when there is interaction between the forward and reverse supply chain in terms of distribution (forward and reverse logistics) and production (new and recovered; see Fig. 16.3). As a simple illustration, we present the following model. Product returns rc can be collected from customer sites c by hybrid collection/recovery centers. There, the products are recovered to ‘as good as new’ condition and can be used to satisfy the demand for products Dc at each customer site c. Because the product

366

E. A. van der Laan

Fig. 16.3 Integrated network combining forward and reverse logistics and combining new manufacturing with recovery operations

returns are not sufficient to cover total demand for products, the brand owner also ships newly manufactured products to the hybrid facilities to be distributed to the customer base. The objective is to optimally choose the forward flows X mi (with m ) between the manufacturing facilities m and hybrid centers i, the transport cost tmi forward flows Yic (with transport cost ticn ) between the hybrid centers i and the customer sites c with used products flow rc , and to optimally assign customer sites to hybrid centers (Z ci = 1 if center i is assigned to customer site c and 0 otherwise; Mi = 1 if hybrid center i is opened, otherwise 0) taking into account the reverse shipping cost tcir from customer site c to hybrid center i and the fixed cost K i of opening hybrid center i. The assignment is constrained by the limited capacity Ci of center i. The total cost function consists of total forward and reverse transport costs plus the costs to open the hybrid centers, as shown next.

16 Closed Loop Supply Chain Management

Minimize

 m

i

m + X mi tmi

 i

c

n + Yic tic

367

 i

c

Z ci tcir rc +



K i Mi

i

subject to the constraints  c 

rc Z ci + Yic =



m Dc

X mi =

i

 c



rc Z ci ≤ Mi Ci Z ci = 1

i

 c

Z ci ≤ Mi N

X mi , Yic ≥ 0 Z ci , Mi ∈ {0, 1}



Yic for all i (supply of new and recovered equals outflow)

c

for all c (each customer site receives the demanded products) for all i (the inflow of returned products at collection center i cannot exceed capacity Ci ) for all j (each customer base is assigned to exactly one collection center) for all i (with N the number of customer sites; now Mi = 1 only if at least 1 customer site is assigned) for all m, i, c (non-negativity) for all i, j (binary variables)

This model is easily extended to incorporate capacities on routes, multiple recovery options and imperfect recovery yields. In Sect. 16.4.1 we look at more realistic, but also more complex models that integrate operational decisions, such as inventory management and pricing.

16.3.3 Combined Manufacturing and Product Recovery When serviceable inventory can be sourced from both the manufacturing of new products and the remanufacturing of used products, we have the situation as sketched in Fig. 16.4. Here we have to balance the manufacturing output with the recovery output to satisfy demands.

Fig. 16.4 Joint manufacturing and recovery system for a remanufacturable product

368

E. A. van der Laan

Under the assumption of deterministic returns and demands, the scheduling model of Sect. 16.2.2 can be extended to allow for new production z t in period t (Richter and Sombrutzki 2000): minimize

T 

rt δ(xt )+

st δ(z t ) + h t yt + Ht It

t=1

subject to the constraints y0 = I0 = 0 yt = yt−1 − xt + dt

for t = 1, . . . , T (balance equation for used product inventory) It = It−1 + xt + z t − Dt for t = 1, . . . , T (balance equation for recovered inventory) for t = 1, . . . , T (non-negativity) It , xt , yt , z t ≥ 0

Here, the constraints to ensure that the number of available products is sufficient to satisfy demand are dropped, as new production can serve as a backup. When flows are stochastic, optimization becomes more complex. Inderfurth (1997) shows that if lead times for recovery and manufacturing are equal and there are no fixed setup costs, the optimal production and inventory policy is still tractable and relatively simple. For each period, it is characterized by a manufacture-up-to level L, a remanufacture-up-to level M and a dispose-down-to level U (a so-called LMU policy, first derived by Simpson (1978)). When this is not the case, the optimal policy is very complex. Fleischmann and Kuik (2003) prove that when fixed manufacturing costs are taken into account, recoverable products are not allowed to be stocked and the fixed recovery lead time is smaller than the manufacturing lead time, a standard (s, S) policy is optimal, that is reorder up to S when the inventory position hits reorder level s. In that case we could approximate the reorder level by the manufacturing lead time net demand plus the safetystock, or (Expected lead time net demand) + k × (Standard deviation of lead time net demand)

(16.2) where k is a safety factor set by management. When recoverable products are allowed to be stocked, a complex trade-off needs to be made between stocking costs, fixed recovery costs and costs of stockouts. To counter this complexity, van der Laan et al. (1999) introduce two types of heuristic policies. The push policy generates an order for a remanufacturing batch as soon as Q r used products are collected. Manufacturing of Q m products is initiated as soon as the serviceable inventory position (new and recovered products plus everything in the production pipeline) reaches some reorder level sm . The pull policy generates an order for a remanufacturing batch of size Q r as soon as the serviceable inventory position reaches reorder level sr and at least Q r recoverable products are available in inventory. Otherwise, manufacturing of Q m products is initiated when the serviceable

16 Closed Loop Supply Chain Management

369

inventory position reaches reorder level sm ≤ sr . In this context, van der Laan and Teunter (2006) develop the following simple manufacturing and remanufacturing order quantities assuming flows are deterministic.  Qm =

2K m (λ − γ ) ; Qr = hs



2K r γ h r + (γ /λ)h s

(16.3)

under push control and  Qm =

2K m (λ − γ ) ; Qr = (γ /λ)h r + (1 − γ /λ)h s



2K r γ h r + (γ /λ)h s

(16.4)

under pull control, with λ the demand rate, γ the return rate, h s the holding cost rate for serviceable products, h r the holding cost rate for recoverable products. These formulas are quite appealing, as they resemble simple EOQ (economic order quantity) type formulas and are shown to be quite robust when flows are stochastic. Let furthermore b denote the backorder cost per occurrence. In the same paper they develop the reorder point formula     hs Qm s = F −1 1 − b λ−γ

(16.5)

for the push policy and ⎛



s = F −1 ⎝1 −

b

hs λ−γ Qm

+

γ Qr



(16.6)

for the pull policy assuming that s = sm = sr . The performance of push and pull type policies is mainly driven by the ratio of recoverable and serviceable holding cost rates. If stocking of recoverables is cheap, then pull control has its advantages; if stocking is expensive, then one should consider push as the better alternative. Other factors that drive towards push control are high backorder costs, long or uncertain recovery lead times, and recovery yield uncertainty. Until now we assumed that manufactured and recovered products are indistinguishable: they are perfect substitutes. This is not necessarily the case in practice. In some cases, remanufactured products are only used as substitutes for manufactured products, for instance as a warranty replacement. In other cases, manufactured products are only used as substitutes for remanufactured products, for instance as emergency supply when a remanufactured product is not available, or to serve as a spare part. For the former case Bayındır et al. (2005, 2007) study the conditions under which substitution is profitable.

370

E. A. van der Laan

16.4 Flow Integration and Data Management (State-of-the-Art) In this section, we briefly review some recent research results and possibilities to reduce uncertainty by exploiting advanced sensor information. Subsequently, we discuss the integration of forward and reverse flows, lead time demand estimation and the impact of the Internet of Things on data management.

16.4.1 Integrated Forward and Reverse Network Design The most advanced network design models not only consider the interaction between the forward and the reverse chain, but also integrate the flow optimization and facility location decisions with operational management decisions. The models in extant literature are too elaborate to present here, but we briefly review them in a qualitative manner. Abdallah et al. (2012) introduce a model that considers location inventory decisions in the forward and reverse supply chain. Distribution centers distribute a product to the retailers facing uncertain demand. The retailers also collect and sort returned products for input to the remanufacturing centers. Decisions are taken regarding the selection of distribution and remanufacturing sites, and assigning retailers to distribution sites and remanufacturing sites. Total costs, consisting of location costs, delivery costs and inventory related costs, are optimized. Asl-Najafi et al. (2015) consider a dynamic, mixed-integer, nonlinear model that integrates distribution and rework facility location with inventory management decisions taking into account disruption risk at the facilities. Decisions to be optimized are selection of distribution and recovery sites and assigning them to retailers, and routes via multiple transportation modes. The bi-objective function to be minimized consists of total costs (the sum of location costs, shipping costs, inventory costs, waste disposal costs and recovery costs) and the total travel time through the network. Kaya and Urek (2016) consider a mixed-integer, nonlinear facility location model that integrates operational decisions. Variables are inventory cycle times, prices for new products and customer incentives to return used products. Total profits to be maximized are calculated as net unit profits of selling new and recovered production minus incentive costs and minus the costs of opening and operating the facilities and inventory costs. The large complexity of the above models prohibits the finding of optimal solutions for practically meaningful instances with standard solver software. Hence, authors resort to the development of heuristics.

16 Closed Loop Supply Chain Management

371

16.4.2 Estimating Lead Time Net Demand In Sect. 16.3.1 we saw that knowing the return distribution allows one to simply calculate various life cycle parameters, such as trippage and average time to return. For inventory management and production planning purposes one often needs to estimate the expectation and standard deviation of the number of returns over some lead time L, or rather the net demand during the lead time (see for instance Eq. 16.2). If we only have an estimate of the mean (μ) and variance (σ 2 ) of the number of demands per time unit and the fraction of products that return (P), one could simply use (Kelle and Silver 1989) E(ND L ) = (1 − P)Lμ, Var(ND L ) = (1 − P)2 Lσ 2 + P(1 − P)Lμ

(16.7)

assuming that demands and returns during the lead time are perfectly correlated. Alternatively one could assume that demands and returns during the lead time are completely uncorrelated and arrive at E(ND L ) = (1 − P)Lμ, Var(ND L ) = (1 + P 2 )Lσ 2 ,

(16.8)

which is a much better approach if P is not too close to one (de Brito and van der Laan 2009). If historical aggregated demand and return information and an estimate of the return time distribution were available (see Sect. 16.3.1), the lead time net demand estimates could be improved significantly by using E(ND L ) = L μ −

t  i=t−n+1

x i Ri − μ

t+L−1 

Ri

(16.9)

i=t+1

where the first term is the expected demand during the lead time, the second term the expected returns from past sales and the third term the returns from sales during the lead time. Here, Ri is the probability of having products from past sales xi returning during the lead time, which can be calculated from the pi ’s. Similarly, we can derive an expression for the variance (Kelle and Silver 1989). If data on historical individual returns and demands were available and each return could be matched with its original sale, we can do even better: E D (ND L ) = L μ −

t−1  i=t−n+1

(u i − z i )Q i − u t Rt − μ

t+L−1 

Ri

(16.10)

i=t+1

Here we make use of conditional probabilities Q i , which reflect the probability that products return from past sales xi during the lead time given that a certain number of products has already returned (Kelle and Silver 1989). Naturally, because the latter expression for estimating the expected lead time net demands uses all of the available information, it should perform better than all of the

372

E. A. van der Laan

other alternatives. However, one should note that that is only under the assumption that we have a perfect estimate for the return time probabilities pi . Toktay et al. (2000) showed that the performance deteriorates when these estimates are inaccurate, where overestimation of the return probabilities is more costly than the other way around. This was confirmed by de Brito and van der Laan (2009), who also showed that even for mild inaccuracies, the most informed method no longer outperforms the method that only uses aggregate data.

16.4.3 Using Sensor Data and the Internet of Things As should be clear by now, the planning and control of closed loop supply chain processes is hampered and complicated due to the high uncertainty in quantity, quality and timing of product returns. Making data loggers part of the design of recoverable products already helps in reducing quality uncertainties upon return. As an example, an early study by Klausner and Hendrickson (1998) showed that the recording of basic usage data in power tools, such as use duration, the number of motor starts and motor temperature, allows the remanufacturing facility to distinguish between recoverable and non-recoverable power tools immediately upon their return instead of during the recovery process. As an added benefit, this data also contributes to product design improvements as the gathered data reveals important information on customer use patterns and preferences, and product reliability (informational value, see Sect. 16.1.3). Another example concerns the tire retreading business. A chip in the tire can extract vital information about the quality of the tire through the recording of tire temperature and number of tire revolutions. A drawback of the technology in the above examples, though, is that the information only becomes available when the product is physically inspected. In the former example information is only available at the end of use; in the latter example it may be available at regular preventive inspections. The Internet of Things makes it possible that remote sensors communicate with central networks or even with other products. This allows one to continuously monitor the installed base and take action when needed. An action can be take back for repair, take back for repair or recycling, send a maintenance team to the customer site or signal the customer that an action is needed. Such technology has tremendous impact on the operational management of the installed base, as no reactive or preventive product replacement is necessary because product failures are spotted or prevented before they happen. Also, the amount of uncertainty in the reverse logistics netork is significantly reduced as one has a much better view on how many products come back when and in what condition. Well before products actually come back, one can already plan capacities of logistics, recovery and manufacturing processes.

16 Closed Loop Supply Chain Management

373

16.5 Further Reading This chapter provides an overview of concepts and modeling efforts within the field of closed loop supply chain management. This is not meant as a comprehensive review of the logistics and operations management literature within the field. Some recommended recent review papers are Govindan et al. (2015), Agrawal et al. (2015a), Vahabzadeh and Yusuff (2015). Instead, we hope that this chapter can inspire newcomers to the field and specialists who would like to broaden their view. A single chapter simply does not allow one to include all relevant concepts and issues. For instance, the vehicle routing part of the reverse logistics network is not covered, although this is an important subproblem to optimize logistics costs for which some studies exist. The interested reader is referred to Soysal (2016) for a recent contribution to this field. Nor did we go into Third Party Reverse Logistic Provider (3PRLP) selection, which can affect quality and price of products, see Govindan et al. (2015). For a recent account of the intracacies of selecting logistics providers in a CLSC context, see Guarnieri et al. (2015). Closed loop supply chains potentially contribute to mitigating environmental damage and risks and enhancing social value, but most studies focus solely on direct economic consequences such as operational costs and profits from sales. Hence, there seems to be ample opportunity to develop models that simultaneously handle economic, environmental and social criteria. One of the few efforts that considers all three criteria is a study by Mota et al. (2015). Applying their network design model to a Portuguese battery recycler reveals that there exist solutions that improve all three criteria when compared to the original situation. We briefly touched upon remote sensing and the Internet of Things to give a flavor of the immense possibilities to reduce uncertainties in return quantities, quality and timing, but more research is needed to explore the exact consequences. A direct consequence from a modeling perspective is that there will be less need for stochastic modeling as deterministic planning models will become more useful in practice. For a complete overview of relevant concepts, frameworks and ideas in this exciting area we refer the interested reader to Kiritsis (2011). Acknowledgements Eliane Haseth kindly provided input for the Gazelle and Volkswagen cases.

References Abdallah T, Diabat A, Simchi-Levi D (2012) Sustainable supply chain design: a closed-loop formulation and sensitivity analysis. Prod Plan Control 23(2–3):120–133 Agrawal S, Singh RK, Murtaza Q (2015a) A literature review and perspectives in reverse logistics. Resour Conserv Recycl 97:76–92 Agrawal VV, Atasu A, Van Ittersum K (2015b) Remanufacturing, third-party competition, and consumers’ perceived value of new products. Manag Sci 61(1):60–72 Aras N, Boyaci T, Verter V (2004) The effect of categorizing returned products in remanufacturing. IIE Trans 36(4):319–331

374

E. A. van der Laan

Aras N, Verter V, Boyaci T (2006) Coordination and priority decisions in hybrid manufacturing/remanufacturing systems. Prod Oper Manag 15(4):528–543 Asl-Najafi J, Zahiri B, Bozorgi-Amiri A, Taheri-Moghaddam A (2015) A dynamic closed-loop location-inventory problem under disruption risk. Comput Ind Eng 90:414–428 Atasu A, Sarvary M, Van Wassenhove LN (2008) Remanufacturing as a marketing strategy. Manag Sci 54(10):1731–1746 Bayındır ZP, Erkip N, Güllü R (2005) Assessing the benefits of remanufacturing option under one-way substitution. J Oper Res Soc 56:286–296 Bayındır ZP, Erkip N, Güllü R (2007) Assessing the benefits of remanufacturing option under one-way substitution and capacity constraint. Comput Oper Res 34:487–514 Blackburn JD, Guide VDR Jr, Souza GC, Van Wassenhove LN (2004) Reverse supply chains for commercial returns. Calif Manag Rev 46(2):6–22 Çorbacıo˘glu U, van der Laan EA (2013) A quality framework in closed loop supply chains: opportunities for value creation. In: Nikolaidis Y (ed) Quality management in reverse logistics. SpringerVerlag London Das K, Chowdhury AH (2012) Designing a reverse logistics network for optimal collection, recovery and quality-based product-mix planning. Int J Prod Econ 135(2):209–221 de Brito MP, Dekker R (2004) A framework for reverse logistics. In: Dekker R, Fleischmann M, Inderfurth K, Van Van Wassenhove LN (eds) Reverse logistics: quantitative models in closed-loop supply chains. Springer de Brito MP, van der Laan EA (2009) Inventory control with product returns: the impact of imperfect information. Eur J Oper Res 194:85–101 Dekker R, van der Laan E (2003) Inventory control in reverse logistics. In: Guide VDR, Van Wassenhove LN (eds) Business aspects of closed-loop supply chains: exploring the issues. Carnegie Mellon University Press Ferguson M, Toktay B (2006) The effect of competition on recovery strategies. Prod Oper Manag 15(3):351–368 Ferrer G, Swaminathan JM (2006) Managing new and remanufactured products. Manag Sci 52(1):15–26 Fleischmann M (2003) Reverse logistics network structures and design. In: Guide VDR, Van Wassenhove LN (eds) Business aspects of closed-loop supply chains: exploring the issues. Carnegie Mellon University Press Fleischmann M, Kuik R (2002) Controlling inventories with stochastic item returns: a basic model. Eur J Oper Res 138:63–75 Fleischmann M, Kuik R (2003) On optimal inventory control with independent stochastic item returns. Eur J Oper Res 151(1):25–37 Goh TN, Varaprasad N (1986) A statistical methodology for the analysis of the life-cycle of reusable containers. IEE Trans 18:42–47 Govindan K, Soleimani H, Kannan D (2015) Reverse logistics and closed-loop supply chain: a comprehensive review to explore the future. Eur J Oper Res 240:603–626 Guarnieri P, Amorim Sobreiro V, Seido Nagano M, Marques Serrano AL (2015) The challenge of selecting and evaluating third-party reverse logistics providers in a multicriteria perspective: a Brazilian case. J Clean Prod 96:209–219 Inderfurth K (1997) Simple optimal replenishment and disposal policies for a product recovery system with leadtimes. OR Spektrum 19(2):111–122 Inderfurth K, Teunter RH (2003) Production planning and control of closed loop supply chains. In: Guide VDR, Van Wassenhove LN (eds) Business aspects of closed-loop supply chains: exploring the issues. Carnegie Mellon University Press Kaya O, Urek B (2016) A mixed integer nonlinear programming model and heuristic solutions for location, inventory and pricing decisions in a closed loop supply chain. Comput Oper Res 65:93–103 Kelle P, Silver EA (1989) Forecasting the returns of reusable containers. J Oper Manag 8(1):17–35

16 Closed Loop Supply Chain Management

375

Kiritsis D (2011) Closed-loop PLM for intelligent products in the era of the Internet of things. Comput Aided Design 43:479501 Kirsner S (2014) Market burgeons for pre-owned smartphones. Boston Globe. https:// www.bostonglobe.com/business/2014/07/05/emerging-market-for-pre-owned-certifiedsmartphones/fFJPIs0AfTymQw1MhrgMYP/story.html Klausner M, Hendrickson CT (1998) Reuse of electric motors in consumer products—design and analysis of an electronic data log. Interfaces 30:156–165 Loomba APS, Nakashima K (2012) Enhancing value in reverse supply chains by sorting before product recovery. Prod Plan Control 23(2–3):205–215 Lyons Hardcastle J (2014) Gazelle pays $200 m for used electronic devices. Environ Lead. https:// www.environmentalleader.com/2014/11/gazelle-pays-200m-for-used-electronic-devices/ Marx-Gomez J, Rautenstrauch C, Nurnberger A, Kruse R (2002) Neuro-fuzzy approach to forecast returns of scrapped products to recycling and remanufacturing. Knowl Based Syst 15(1–2):119– 128 Matsumoto M, Komatsu S (2015) Demand forecasting for production planning in remanufacturing. Int J Adv Manuf Technol 79:161–175 Minner S, Kleber R (2001) Optimal control of production and remanufacturing in a simple recovery model with linear cost functions. Or Spektrum 23(1):3–24 Mota B, Gomes MI, Carvalho A, Barbosa-Povoa AP (2015) Towards supply chain sustainability: economic, environmental and social design and planning. J Clean Prod 105:14–27 Muckstadt JA, Isaac MH (1981) An analysis of single item inventory system with returns. Naval Res Logistics Q 28:237–254 Ovchinnikov A (2011) Revenue and cost management for remanufactured products. Prod Oper Manag 20(6):824–840 Richter K (1996) The EOQ repair and waste disposal model with variable setup numbers. Eur J Oper Res 95(11):1412–1426 Richter K, Sombrutzki M (2000) Remanufacturing planning for the reverse Wagner/Whitin models. Eur J Oper Res 121(2):304–315 Schenkel M, Caniëls MCJ, Krikke H, van der Laan E (2015) Understanding value creation in closed loop supply chains Past findings and future directions. J Manuf Syst 37:729–745 Schrady DA (1967) A deterministic inventory model for repairable items. Naval Res Logistics Q 14:391–398 Simpson VP (1978) Optimum solution structure for a repairable inventory problem. Oper Res 26(2):270–281 Soysal M (2016) Closed-loop inventory routing problem for returnable transport items. Trans Res Part D: Trans Environ 48:31–45 Tagaras G, Zikopoulos C (2008) Optimal location and value of timely sorting of used items in a remanufacturing supply chain with multiple collection sites. Int. J. Prod Econ 115:424–432 Teunter RH (2001) Economic ordering quantities for recoverable item inventory systems. Naval Res Logistics 48(6):484–495 Toktay LB, Wein LM, Zenios SA (2000) Inventory management of remanufacturable products. Manag Sci 46(11):1412–1426 Vahabzadeh AH, Yusuff RBM (2015) A content analysis in reverse logistics: a review. J Stat Manag Syst 18(4):329–379 van der Laan E, Salomon M, Dekker R, Van Wassenhove L (1999) Inventory control in hybrid systems with remanufacturing. Manag Sci 45(5):733–747 van der Laan EA, Teunter RH (2006) Simple heuristics for push and pull remanufacturing policies. Eur J Oper Res 175:10841102

Part V

Models for Operations, Logistics and Supply Chain Management

Chapter 17

Location Analysis and Network Design Mark S. Daskin and Kayse Lee Maass

Abstract This chapter begins with a basic taxonomy of facility location models. This is followed by the formulation of five classic facility location models: the set covering model, the maximum covering model, the p-median model, the fixed charge location model and the p-center problem. Advanced: Computational results on a new set-covering problem instance with 880 nodes representing 880 population centers in the contiguous United States are provided and a few counter-intuitive results are outlined. This is followed by a state of the art discussion of multi-objective problems in location analysis and the importance of multiple objectives in designing distribution networks. Models that integrate inventory planning into facility location modeling are then outlined. Finally, the chapter ends with a discussion of reliability in facility network planning.

17.1 Introduction (Basic) The long-term performance capabilities of a supply chain are determined largely by the location of the components of the supply chain. For example, the ability of a company like Amazon to provide timely deliveries to its customers depends in large part on the location of its warehouses relative to its customer base and on the assortment of goods at each warehouse. In the case of automobile assembly plants, the price of automobile components purchased overseas may be less than the purchase price of similar products made domestically, but the total cost of such goods may be higher after the transport costs are included in the price. Similarly, the minimum cost configuration of suppliers in the US may call for three suppliers—one on the

M. S. Daskin (B) Industrial and Operations Engineering, University of Michigan, Ann Arbor, MI, USA e-mail: [email protected] K. L. Maass Mechanical and Industrial Engineering Department, Northeastern University, Boston, MA, USA e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_17

379

380

M. S. Daskin and K. L. Maass

west coast, one on the east coast, and one in the central part of the country. However, when supplier failures are considered, additional suppliers may prove beneficial. The design of a supply chain is intricately linked to the location of the facilities in the system. Facility locations represent long-term decisions; it is often either impossible or very costly to change locations frequently over the lifetime of the supply chain. Thus, locations that are not aligned with and that do not support the strategic goals of the company may lead to long-term persistent inefficiencies and/or an inability to achieve the firm’s strategic goals. Consequently, this chapter is concerned with facility location models. The key decisions associated with facility location modeling are: 1. 2. 3. 4.

How many facilities should be used in the supply chain? Where should the facilities be located? How should demand nodes be allocated to facilities? What is the level of service provided to the customers of the supply chain with a given configuration? In addition, some facility location models are designed to answer: 5. How large should each facility be? 6. What plans are there for backup facilities in the event of facility failures? To motivate the study of location problems, consider the issues faced by Amazon.com.

17.2 Case Study: Amazon.Com Amazon.com is the world’s largest internet-based retailer today. However, it is easy to forget that it started only 22 years ago, in 1994, as an online bookseller. Today, one can buy virtually anything on Amazon.com except for cars and homes. As Amazon enters new businesses, it needs to determine where to locate warehouses, how to allocate customers to those warehouses, which products to stock at each warehouse, how to ship products to customers, and what level of service it wants to provide to each customer class. A customer class may be determined by the customer’s geographic location, by his/her past purchasing behavior, and by the customer’s relationship with the company (e.g., whether or not the customer is an Amazon Prime customer). The company clearly wants to provide service as efficiently and effectively as possible, yet recognizes that not all customers experience the same service levels. As a result, the company is concerned with the range of customer experiences it provides and its customers perceive. Twenty-two years after its founding, the company has approximately 240 fulfillment and distribution centers, sortation facilities, delivery stations and Amazon Prime locations encompassing over 92 million square feet of space in the United States alone. Globally, Amazon has over 390 facilities with over 140 million square feet of space in its various facilities. Many of these facilities

17 Location Analysis and Network Design

381

are specialized for particular types of products (e.g., perishable and frozen food products.) Amazon does not publicly disclose the size and location of its facilities, but a good publicly available source of information on the Amazon network can be found at (Amazon 2017). Clearly, with such an expansive network of customers, the number, type and location of facilities is critical to Amazon’s ability to remain a leading online retailer.

17.3 Taxonomy of Location Models (Basic) Broadly speaking there are four categories of location models (Daskin 2013): 1. Analytic models make very strong assumptions about the underlying data. For example, we may assume that the demands are uniformly distributed over the service region and that facilities can be located anywhere in the service region. Additional assumptions are typically made regarding the direction(s) of travel within the region; e.g., we may assume that travel occurs along horizontal or vertical lines only or that travel occurs in a straight line between any two points in the region. Such models can generally be solved exactly in closed form. Analytic models are useful for insights into the structure of the problem and its solution, but are rarely useful in determining where exactly to locate facilities. For example, under appropriate assumptions (Daskin 2013), the annual transport costs should be twice the annualized facility costs. In practice, real systems deviate significantly from this idealized result; often the facility location and transport costs are roughly equal. 2. Continuous location models assume that facilities can be located anywhere in the plane. The facilities serve demands of different magnitudes that occur at discrete locations in the plane. The Weber problem is an example of such a location problem (Drezner et al. 2001). In this model, we seek the location of a single facility to minimize the demand-weighted total distance between the facility and the demand nodes. The demand weighted total distance is the sum (over all demand nodes) of the product of the demand at each node multiplied by the distance between that node and the facility. With a single facility to be located, this problem can be solved numerically very effectively. When two or more facilities are to be located, we need to simultaneously decide how to allocate demand to the facilities and where they should be. This problem is considerably more difficult and multiple optima are likely to exist. 3. Network location models entail locating facilities on a network of nodes and arcs. Much of the literature on network location models focuses on finding polynomial time algorithms for specially structured instances of various problems. For example, many problems are structured on trees. A tree is a network that consists of a set of undirected edges with exactly one way of going between each pair

382

M. S. Daskin and K. L. Maass

of nodes on the network. Locating a single facility on a tree to minimize the demand weighted total distance can be done in an amount of time that is linear in the number of nodes (Goldman 1971). The optimal location is on a node such that half or more of the demand is at that node or to one side of that node. Indeed, polynomial time algorithms exist for this problem on a tree for any number of facilities, p, and any number of demand nodes, n (Kariv and Hakimi 1979). 4. Finally, discrete location models do not necessarily assume that there is an underlying network. Instead, there typically is a discrete set of candidate demand nodes (e.g, the centroids of the 3,109 counties of the contiguous United States) and a discrete set of candidate nodes (e.g., the centroids of the 100 most populous counties in the contiguous United States). The focus of this chapter is on discrete location models.

17.4 Basic Discrete Model Forms (Basic) Within the class of discrete location models, there are three broad classes of location models: covering models, median-based models, and a variety of other models. A demand node is (generally) said to be covered if the distance or time to the nearest facility is less than or equal to some specified service or coverage distance or time. The set covering model finds the minimum number or minimum cost of the facilities needed to cover all demand nodes at least once (Toregas et al. 1971). Often the number of facilities needed to cover all demands (or the cost of those facilities) is prohibitively large. In those cases, we limit the number of facilities deployed and solve a maximum covering model. The maximum covering model finds the locations of a specified number of facilities to maximize the number of covered demands. Finally, within the class of covering models, the p-center model finds the locations of p facilities such that all demands are covered and the coverage distance is as small as possible. To formulate the set covering model, we define the following inputs: Inputs: I J di j fj Dc

the set of demand nodes the set of candidate locations the distance between demand node i ∈ I and candidate site j ∈ J the (fixed) cost of locating a facility at candidate site j ∈ J the coverage distance  ai j

and the decision variable

1 if di j ≤ D c 0 otherwise

17 Location Analysis and Network Design

 Xj

383

1 if we locate a facility at candidate site j ∈ J 0 otherwise

With this notation, we can define the location set covering model as follows:  Min (17.1) Xj s.t.



j∈J

ai j X j ≥ 1 ∀i ∈ I

j∈J

X j ∈ {0, 1} ∀ j ∈ J

(17.2) (17.3)

The objective function (17.1) minimizes the number of facilities used. (If we want to minimize  the cost of the located facilities we would replace this objective function with j∈J f j X j .) Constraints (17.2) stipulate that each demand node should be covered by at least one facility. Finally, constraints (17.3) ensure that the location variables can only take on values of 0 or 1. While the set covering problem is NP-hard, in practice large instances of the problem can be solved to optimality. Throughout the remainder of this chapter, we will present computational results for an 880-node problem instance, which represents the 880 zip-3 regions of the contiguous United States. The dataset was developed by aggregating the five-digit zip code data for the 2010 census into the corresponding 3-digit zip code regions. The most populous node corresponds to the 770 zip-3 region in Houston, Texas with over 2.9 million people. Seven regions have over 2 million people and 55 regions have over 1 million people. The total population in the dataset is 306,669,700. The distance between nodes is the great circle distance rounded to the nearest mile. Figure 17.1 is a map of the 880 nodes. The seven green squares correspond to locations with a population of over 2 million people. The vertical lines are proportional to the demand at each location. Figure 17.2 shows the results for the set covering model for this dataset. Clearly as the coverage distance increases, the number of needed facilities decreases. For this dataset, a power regression suggests the number decreases as the 1.73 power (R 2 value of 0.999) of the coverage distance. There are a number of significant problems with the set covering problem including: (a) the number of facilities needed to cover all demands often exceeds the number than can be deployed; (b) the model treats all nodes as being equally important in the sense that both small and large demand nodes must be covered at least once; (c) the model does not provide for backup coverage in the event that a facility is busy (e.g., an ambulance is busy) or that a facility has failed (e.g., due to weather conditions or due to a terrorist attack); (d) there are often alternate optima and the model provides no way of discriminating between these different solutions. Addressing each of these concerns is beyond the scope of this chapter. However, the maximum covering model (Church and ReVelle 1974) addresses the first two concerns outlined above. By introducing the inputs

384

Fig. 17.1 Map of relative demands for the 880-node dataset

Fig. 17.2 Set covering results for the 880-node dataset

h i the demand at node i ∈ I p the number of facilities to locate

M. S. Daskin and K. L. Maass

17 Location Analysis and Network Design

385

and decision variable  Zi

1 if demand node i ∈ I is covered 0 otherwise

to the previously defined notation, we can formulate the maximum covering problem as follows:  hi Zi (17.4) Max s.t. Z i −



i∈I

ai j X j ≤ 0 ∀i ∈ I

(17.5)

Xj  p

(17.6)

j∈J

 j∈J

X j ∈ {0, 1} ∀ j ∈ J

(17.7)

Z i ∈ {0, 1} ∀i ∈ I

(17.8)

The objective function (17.4) maximizes the number of covered demands. Constraints (17.5) link the location and coverage variables. They stipulate that demand node i ∈ I can only be counted as being covered  (Z i  1) if there is at least one chosen facility that can cover the demand node ( j∈J ai j X j ≥ 1). Constraint (17.6) states that we are to locate p facilities. Constraints (17.7) and (17.8) are integrality constraints. There are numerous heuristics for the maximum covering model. However, reasonably large instances of the problem can be solved optimally using either Lagrangian relaxation (relaxing constraint (17.6)) and embedding the Lagrangian relaxation in branch and bound or by using commercially available solvers. Figure 17.3 shows the results for the maximum covering model using the 880-node dataset with coverage distances of 200, 300, 400 and 500 miles. For these distances, 31, 16, 10 and 7 facilities are needed respectively to cover all demands. However, only 15, 9, 6 and 4 facilities are needed to cover 90% of the total demand for these four distances respectively. For many datasets, 80–90% of the total demand can be covered with half of the number of facilities needed to cover the entire demand. Figure 17.4 shows the optimal maximum covering solution using 9 facilities and a coverage distance of 300 miles. Each demand node is connected to the nearest facility. Red connections are within 300 miles and blue connections indicate distances greater than the 300-mile coverage distance. For any coverage distance, the percent of the total demand that is covered increases as additional facilities are added, as shown in Fig. 17.3. Generally, the rate of increase is decreasing, meaning that each additional facility that is added increases the coverage by less than did the preceding facility. This is not always the case, however. Table 17.1 shows the total coverage and the improvement in coverage as we go from 19 to 22 facilities with a coverage distance of 200 miles. The coverage increases by 2,450,935 as the 20th facility is added and by 1,888,488 as the 21st facility is added.

386

M. S. Daskin and K. L. Maass

Fig. 17.3 Maximum covering results for the 880-node dataset

Fig. 17.4 Maximum covering solution with 9 facilities and a coverage distance of 300

17 Location Analysis and Network Design

387

Table 17.1 Number of facilities, covered demand, and improvement in coverage for a coverage distance of 200 miles Number of facilities Covered demand Improvement 19

294,728,807

2,450,935

20

297,179,742

1,888,488

21

299,068,230

1,915,151

22

300,983,381

However, the coverage increases by 1,915,151 as the 22nd facility is added; this is more than the increase resulting from the addition of the 21st facility. Other instances of this rare occurrence are documented in Daskin (2013). We defer formulation of the p-center problem, the last major covering model, until after we have outlined the class of median problems. The p-median problem (Hakimi 1965) selects p facilities from a set of candidate nodes to minimize the demand-weighted total distance between demand nodes and the nearest open facility. In addition to the notation outlined above, we define the following additional decision variable:  1 if demand node i ∈ I is assigned to a facility at j ∈ J Yi j 0 if not With this additional notation, the p-median model can be formulated as follows: Min



h i di j Yi j

i∈I j∈J

s.t.



Yi j  1 ∀i ∈ I

j∈J

Yi j ≤ X j ∀i ∈ I ; j ∈ J  Xj  p

(17.9) (17.10) (17.11) (17.12)

j∈J

X j ∈ {0, 1} ∀ j ∈ J

(17.13)

Yi j ∈ {0, 1} ∀i ∈ I ; j ∈ J

(17.14)

The objective function (17.9) minimizes the demand-weighted distance. Constraints (17.10) ensure that each demand node is assigned exactly once. Constraints (17.11) state that demand nodes can only be assigned to open facilities. Constraint (17.12) ensures that we locate exactly p facilities. Constraints (17.13) and (17.14) are standard integrality constraints. Note that constraints (17.14) can be relaxed in this model to simple non-negativity constraints. Figure 17.5 plots the demand-weighted average distance (the objective function (17.9) divided by the total population) versus the number of medians. The distance

388

M. S. Daskin and K. L. Maass

Fig. 17.5 Average distance versus number of medians for 880-node dataset Table 17.2 Selected demand-weighted average distances and improvements for 880-node problem Medians Demand-weighted average Improvement 16 17 18 19 76 77 78 79

112.9755 107.8393 103.3294 98.7628 37.0574 36.6873 36.3359 35.9843

5.1362 4.5098 4.5667 0.3702 0.3514 0.3515

decreases with the 0.71 power of the number of medians. This is faster than the rate predicted by an analytic model (which would predict that distance would decrease as the 0.5 power of the number of medians (Daskin 2010)) due to the non-uniformity of the demands in the real dataset as shown in Fig. 17.1. As additional facilities are added, the demand weighted average or total distance decreases. The rate of decrease is generally decreasing, though this too is not always the case as shown in Table 17.2. For example, when we increase the number of medians from 16 to 17, the average distance decreases by 5.1362 miles. When we add another median, the average goes down by 4.5098 miles. However, when we add yet another (getting to 19 total medians), the average goes down 4.5667 miles, even more than the decrease that resulted from adding the 18th facility. The same phenomenon occurs between 76 and 79 medians as shown in the table. The fixed charge location problem is similar to the p-median problem except that instead of limiting the number of facilities to be located (as is done in constraint (17.12)), the model endogenously determines the number to use by balancing the fixed facility costs against the transportation costs. Letting β represent the cost per

$18,00,000 $16,00,000 $14,00,000 $12,00,000 $10,00,000 $8,00,000 $6,00,000 $4,00,000 $2,00,000 $0 $0

389

Total Cost and Number of Sites vs. Beta

8 6 4 2

$2

$4

$6

$8

$10

$12

Number of FaciliƟes

Total Cost

17 Location Analysis and Network Design

0 $14

Cost per item - mile (x 10^6) Fig. 17.6 Total cost and number of facilities versus cost per item-mile for the 880-node dataset

item-mile, the fixed charge location problem can readily be formulated with the notation defined above as follows: Min



fjXj + β

j∈J



h i di j Yi j

i∈I j∈J

(17.15)

s.t. (17.10), (17.11), (17.13) and (17.14). The objective function (17.15) minimizes the sum of the fixed facility costs and the transportation costs. Clearly, as β increases, the transportation costs become increasingly important and the model will locate additional facilities to offset the increase in transport costs. This is shown in Fig. 17.6 for the 880-node dataset. Finally, if we let Q be a decision variable representing the maximum assigned distance, we can formulate the p-center problem as follows:

s.t.

 j∈J

Min Q di j Yi j ≤ Q ∀i ∈ I

(17.16) (17.17)

and (17.10)–(17.14). The objective function (17.16) minimizes the maximum distance and constraints (17.17) define the maximum distance in terms of the assigned distances for each demand node. In the next section, we explore a key tradeoff between the average and maximum distance metrics.

390

M. S. Daskin and K. L. Maass

17.5 Multiple Objectives and Incorporating Inventory into Location Models (Advanced) Few if any supply chains are designed with a single objective: they usually must satisfy multiple stakeholders with different objectives. For example, Amazon customers want to get their goods as quickly as possible at the least possible cost. The company wants to provide excellent service in terms of delivery time and cost to its customers, but also wants to make a profit on each item sold and delivered. It is also concerned with both the average level of customer service and the service provided to those who receive the worst service. Multiple conflicting objectives are even more dramatic in some public-sector problems. Citizens want emergency services provided as quickly as possible when they are needed. At the same time, people want to pay as little as possible for city services. In this section, we begin by discussing models with multiple objectives. There is often a conflict between the median (average) and center (maximum) objectives. For example, using the 880-node dataset, the solution to the 5-median problem results in an average distance of 250.6 miles and a maximum distance of 1037 miles. If we solve the 5-center problem, we find that the smallest possible maximum distance is 600 miles (more than 42% less than the maximum found for the 5-median solution), but this solution results in an average distance of 332.7 miles (a solution value that is nearly 33% greater than the optimal 5-median solution value). However, these are only two of 52 non-dominated solutions that can be found for this problem instance. Figure 17.7 plots the tradeoff between the average and maximum distance in this case. The red line shows the 12 solutions that could be obtained using the weighting method, while the blue dots show the full set of 52 non-dominated solutions that were obtained using the constraint method (Cohon 1978). Very good alternatives to the extreme median (top leftmost) and center (bottom rightmost) solutions exist. For example, the maximum distance can be reduced by 7.4% to 960 miles with only a 0.3% increase in the average distance to 251.4. Similarly, the average distance can be reduced from that associated with the center problem by 5.5% to 314.3 miles with only a 0.3% increase in the maximum distance to 602 miles. In addition, numerous other good compromise solutions are shown in the figure. Interestingly, only 59 of the 880 nodes are used in these 52 solutions. Three-digit zip code 079 (in New Jersey) is used in 33 solutions; codes 752 and 751 (in Texas) is used in 21 and 15 solutions, respectively; and codes 306 and 312 (in Georgia) are both used in 18 solutions. This tradeoff considers the average and worst-case performance of a set of candidate locations. If there is a service standard (for example, 1-day delivery service), we can explore the tradeoff between the average distance, which is a good proxy for the delivery cost, and the coverage, a measure of customer service. This can be done using either the constraint or weighting method (Cohon 1978). Figure 17.8 plots the tradeoff between the average distance and the percent of the demand that is covered within 250 miles for the 880-node dataset using 5 facilities. If average distance minimization is paramount, only 56.3% of the demand can be covered. On the other

17 Location Analysis and Network Design

391

Fig. 17.7 Tradeoff between average and maximum distance for the 880-node dataset using 5 facilities

Fig. 17.8 Tradeoff between coverage and average distance

hand, if service is most important, 65% of the demand can be covered in 250 miles (a 15.4% improvement), but the average distance increases 5% from 250.6 miles to 264.0 miles. Excellent compromise solutions exist. For example, by increasing the average distance only 0.5% (to 251.9 miles), the percent of demand covered can be increased over 10 percent to 63 percent. Again, it is critical that tradeoffs and multiple objectives be considered when locating facilities for supply chain management problems. Daskin (2013) outlines methods for finding such non-dominated solutions. Just as it is important to consider multiple conflicting objectives, so too is it important to assess other key supply chain operations when determining the locations of facilities in the supply chain. Significant work has been done on incorporating inventory into facility location models. Nozick and Turnquist (2001) introduce a nonlinear joint location inventory model. They solve the problem by “guessing” at the

392

M. S. Daskin and K. L. Maass

location of the solution, linearizing the objective function at that point, determining a new optimal solution based on the linearization, re-linearizing at that point, and so on, until the model converges. Shen (2000), Daskin et al. (2002) and Shen et al. (2003) incorporate a (Q,r) inventory model with type I service (Hopp and Spearman 1996; Nahmias 1997). They solve the non-linear optimization problem using Lagrangian relaxation embedded in branch and bound (Daskin et al. 2002) or by restructuring the problem as a set covering model and using column generation techniques (Shen 2000; Shen et al. 2003). Numerous extensions of these models have been proposed (Ozsen et al. 2008, 2009; Shen and Daskin 2005; Shen and Qi 2007; Shu et al. 2005; Snyder et al. 2007). Finally, another line of incorporating inventory into facility location models aims to mitigate hard capacity constraints that are often incorporated into the fixed charge facility location model. Such constraints take the following form:  i∈J

h i Yi j ≤ k j X j ∀ j ∈ J

(17.18)

where k j is the capacity of a facility at location j ∈ J . Constraint (17.18) limits the total demand that is assigned to a facility at j ∈ J to be less than or equal to the capacity of the facility if we locate there and to 0 if we do not locate there. There are numerous problems with this constraint. First, it is hard to measure the capacity and companies often do so in ways that are simply incorrect (e.g., by measuring the dollar volume of business flowing through a warehouse). Second, operational tactics including overtime, outsourcing, and holding orders for later fulfillment enable managers to adjust the capacity effectively. Third, the demands, h i , represent average demands and fail to capture the stochastic nature of most demand processes as well as temporal correlations in the demands. For example, demands for some products and services may peak on weekends when most people are not working, while the demand for other products and services may exhibit the opposite tendency. Maass et al. (2016) outline a model that uses inventory to mitigate these hard capacity constraints. Maass and Daskin (2017) propose a cyclic allocation policy that begins to account for cyclic demand or supply variations (including time-dependent hours of operation of facilities). It is common in the facility location literature for models to focus on the physical costs of the system, such as those associated with building and operating a facility, transporting demands from customers to facilities, and holding inventory. However, a number of qualitative factors also contribute to location decisions and network design. When locating obnoxious or noxious facilities, such as nuclear reactor plants or waste disposal sites, decision makers must assess the effect of environmental pollution and healthcare implications on nearby population centers (Hosseini and Esfahani 2009). Furthermore, many companies are becoming increasingly cognizant of the economic, environmental, and social dimensions that contribute to their triple bottom line regarding the interplay between profit, people, and the planet (Chen et al. 2014). Among other things, these factors include the political stability (Dou

17 Location Analysis and Network Design

393

and Sarkis 2010), competitiveness (Jorgensen and Knudsen 2006), and workers’ rights standards (Carter et al. 2008) in the region.

17.6 Reliability in Location Modeling (State-of-the-Art) In the aftermath of the terrorist attacks of September 11, 2001, much research has focused on making supply chains more resilient to attacks and more robust. There has been a general acceptance that (1) the inputs to facility location models are inherently uncertain and (2) there will be times when some facilities are not available due to facility disruptions that are caused by either terrorist actions or acts of nature (e.g., blizzards, hurricanes or earthquakes). To motivate the need to consider reliability in facility location decision making, consider the solution to the fixed charge location model with a cost per item-mile of 0.000012. Figure 17.9 plots this solution. If all facilities are working, the cost is $1,552,585. However, if the facility in the 289 3-digit zip code fails (the facility in North Carolina), the cost skyrockets to $2,283,554 due to the need to reallocate demands to more remote facilities. If the probability of any one of these facilities failing is only 0.02 and we ignore the possibility of multiple failures, the expected cost of this solution is $1,589,035. Now consider the solution shown in Fig. 17.10. This solution adds a facility in the 994 3-digit zip code in the state of Washington. The cost when all six facilities are working is $1,557,049 or less than 0.3% more than the cost of the solution shown in Fig. 17.9. However, the maximum failure cost is now only $1,961,336, or more than 14% less than the maximum failure cost associated with the solution in Fig. 17.9. Again, if the probability of any one facility failing is 0.02, the expected cost of this solution is $1,587,449 or $1,586 less than the expected cost of the solution shown in Fig. 17.9. In fact, for any failure probability greater than 0.0148, the expected cost of the six-site solution in Fig. 17.10 is less than that of the five-site solution shown in Fig. 17.9. Figure 17.11 plots the tradeoff between the maximum failure cost and the cost when all facilities are working in the blue line. The leftmost point is the solution to the traditional fixed charge location problem shown in Fig. 17.9. The point at the sharp elbow in the curve corresponds to the solution shown in Fig. 17.10. The 11 solutions shown in Fig. 17.11 were generated using a genetic algorithm (Goldberg 1989; Haupt and Haupt 1998; Mitchell 1996). Genetic algorithms are a particularly effective means of identifying tradeoffs in location modeling since (1) the encoding of solutions is often simple and (2) any population of solutions automatically gives an approximation of the tradeoff curve (though initial generations in a genetic algorithm may not give a good approximation of the true tradeoff curve). If the primary threats to a supply chain are from nature, we can model the problem in a variety of ways. One approach is to assume that all facilities fail with the same probability and that facility failures are independent. Snyder and Daskin (2005) made this assumption, for example. This class of models assigns demands to multiple facilities to account for facility failures. Cui et al. (2010) relax the assumption of

394

M. S. Daskin and K. L. Maass

Fig. 17.9 Solution to fixed charge location problem with the cost per item-mile = 0.000012

identical facility failure probabilities and allow site-specific failure probabilities. Recently, Li et al. (2013) introduced an approach that not only eliminates the need for failure probabilities to be identical, but it also allows failures to be correlated. If the primary threats to a supply chain are man-made, bi-level and tri-level optimization models are appropriate. In a bi-level optimization model, the top level represents the malevolent player (e.g., the terrorist) whose goal is to inflict as much damage as possible on the system. The lower level represents the system operator who must do his or her best to operate the system after the infliction of the damage by the malevolent player. The tri-level optimization generally adds a layer on top of the malevolent player. This level again represents the system operator. At the top level, he or she can harden some facilities so that they are not susceptible to attacks or so that they are less vulnerable to attacks. Church et al. (2004) formulate variants of the p-median and maximum covering models in which they assume they know where the facilities are located, but identify r facilities whose removal would degrade the system the most. Scaparra and Church (2008a) extend the r-interdiction median model to identify a set of q facilities to harden to minimize the impact of an attack by a terrorist capable of removing r of the original p facilities. Church and Scaparra (2007) show that this can be restructured as a variant of a covering problem in which one is covering interdiction plans. This approach is extended in Scaparra and Church

17 Location Analysis and Network Design

Fig. 17.10 Map of a more reliable solution

Fig. 17.11 Maximum failure cost/fixed charge location cost tradeoff

395

396

M. S. Daskin and K. L. Maass

(2008b). Liberatore et al. (2011) relax the assumption that the number of facilities that the malevolent attacker can remove from the system is known a priori. Liberatore et al. (2012) introduce the possibility of correlated damage as the result of attacks.

17.7 Further Reading We refer readers interested in a more detailed overview of discrete facility location models to Daskin (2013). In addition to discussing many of the models described in this chapter in greater detail, Daskin (2013) describes a number of discrete facility location models not covered in this chapter, including hierarchical facility location models, interacting facilities, and hub location problems. Farahani et al. (2010) and Nickel et al. (2015) present recommended literature reviews on multiple objective facility location models; Farahani et al. (2010) provide a detailed classification of the related literature and is recommended for the reader interested in a basic understanding of such models, while Nickel et al. (2015) provide a more mathematically rigorous discussion. To readers interested in integrated supply chain decisions, we suggest Shen (2007), which provides a survey of location-inventory, location-routing, and inventory-routing models. We recommend Snyder et al. (2006) and Snyder et al. (2016) for reviews of location models subject to disruptions. Chan (2005) examines location modeling in the context of land-use modeling; spatial equilibrium modeling; and trip generation, competition and distribution modeling. Chan (2011) includes a discussion of the use of other methodologies including simulation as well as the impact of remote sensing and geographical information systems on location decisionmaking.

References Amazon Global Fulfillment Center Network (2017). http://www.mwpvl.com/html/amazon_com. html. Accessed 2 May 2017 Carter CR, Rogers DS (2008) A framework of sustainable supply chain management: moving toward new theory. Int J Phys Distrib Logist Manag 38(5):360–387 Chan Y (2005) Location, transport and land-use: modelling spatial-temporal information. Springer, Berlin Chan Y (2011) Location theory and decision analysis: analytics of spatial information technology, 2nd edn. Springer, Berlin Chen L, Olhager J, Tang O (2014) Manufacturing facility location and sustainability: a literature review and research agenda. Int J Prod Econ 149:154–163 Church RL, ReVelle C (1974) The maximal covering location problem. Pap Reg Sci Assoc 32:101–118 Church RL, Scaparra MP, Middleton RS (2004) Identifying critical infrastructure: the median and covering facility interdiction problems. Ann Assoc Geogr 94(3):491–502 Church RL, Scaparra MP (2007) Protecting critical assets: the r-interdiction median problem with fortification. Geogr Anal 39(2):129–146

17 Location Analysis and Network Design

397

Cohon JL (1978) Multiobjective programming and planning. Academic Press, New York Cui T, Ouyang Y, Shen ZJM (2010) Reliable facility location design under the risk of disruptions. Oper Res 58:4-part-1, 998–1011 Daskin MS (2010) Service science. Wiley, NY Daskin MS (2013) Network and discrete location: models, algorithms and applications, 2nd edn. Wiley, NY Daskin M, Coullard C, Shen ZJ (2002) An inventory-location model: formulation, solution algorithm and computational results. Ann Oper Res 110:83–106 Dou Y, Sarkis J (2010) A joint location and outsourcing sustainability analysis for a strategic offshoring decision. Int J Prod Res 48(2):567–592 Drezner Z, Klamroth K, Schöbel A, Wesolowsky GO (2001) The weber problem. In: Drezner Z, Hamacher HW (eds) Facility location: applications and theory, Springer, Heidelberg, pp. 1–36 Farahani RZ, SteadieSeifi M, Asgari N (2010) Multiple criteria facility location problems: a survey. Appl Math Model 34:1689–1709 Goldberg DE (1989) Genetic algorithms in search, optimization and machine learning. AddisonWesley, Reading, MA Goldman AJ (1971) Optimal center location in simple networks. Transp Sci 5:212–221 Hakimi SL (1965) Optimum distribution of switching centers in a communication network and some related graph theoretic problems. Oper Res 13:462–475 Haupt RL, Haupt SE (1998) Practical genetic algorithms. Wiley, New York, NY Hopp W, Spearman ML (1996) Factory physics: foundations of manufacturing management. Irwin, Chicago Hosseini S, Esfahani AM (2009) Obnoxious facility location. In: Farahani, RZ Hekatfar M (eds) Facility location: concepts, models, algorithms and case studies, Springer, pp 315–345 Kariv O, Hakimi SL (1979) An algorithmic approach to network location problems. II: The pmedians. SIAM J Appl Math 37:539–560 Li X, Ouyang Y, Peng F (2013) A supporting station model for reliable infrastructure location design under interdependent disruptions. Transp Res Part E 60:80–93 Liberatore F, Scaparra MP, Daskin MS (2011) Analysis of facility protection strategies against an uncertain number of attacks: the stochastic r-interdiction median problem with fortification. Comput Oper Res 38(1):357–366 Liberatore F, Scaparra MP, Daskin MS (2012) Hedging against disruptions with ripple effects in location analysis. Omega 40(1):21–30 Maass KL, Daskin MS, Shen S (2016) Mitigating hard capacity constraints with inventory in facility location modeling. IIE Trans 48(2):120–133 Maass KL Daskin MS (2017) A cyclic allocation model for the inventory-modulated capacitated location problem. Inf Syst Oper Res Adv. Online Publication Nahmias S (1997) Production and operations management, 3rd edn. Irwin, Chicago Nickel S, Puerto J, Rodríguez-Chía AM (2015) Location problems with multiple criteria. In: Laporte G, Nickel S, Saldanha da Gama F (eds) Location science, Springer, pp. 205–247 Nozick, LK, Turnquist, MA (2001) A two-echelon inventory allocation and distribution center location analysis. Transp Res E 37:425–441 Ozsen, L, Coullard, C, Daskin, MS (2008) Capacitated facility location model with risk pooling. Naval Res Logistics 55(4):295–312 Ozsen, L, Coullard, C, Daskin, MS (2009) Facility location modeling and inventory management with multi-sourcing. Transp Sci 43(4):455–472 Toregas C, Swain R, ReVelle C, Bergman L (1971) The location of emergency service facilities. Oper Res 19:1363–1373 Scaparra MP, Church RL (2008a) A Bilevel Mixed-integer program for critical infrastructure protection planning. Comput Oper Res 35(6):1905–1923 Scaparra MP, Church RL (2008b) An exact solution approach for the interdiction median problem with fortification. Eur J Oper Res 189(1):76–92

398

M. S. Daskin and K. L. Maass

Shen, Z-JM (2000) Efficient Algorithms for various supply chain problems. PhD dissertation, industrial engineering and management sciences, Northwestern University, Evanston, IL, USA Shen, Z-JM, Coullard, C, Daskin, MS (2003) A joint location-inventory model. Transp Sci 37(1):40–55 Shen, Z-JM, Qi, L (2007) Incorporating inventory and routing costs in strategic location models. Eur J Oper Res 179(2):372–389 Shen ZM (2007) Integrated supply chain design models: a survey and future research directions. J Ind Manag Optim 3(1):1–27 Shu, J, Teo, C-P, Shen, Z-JM. (2005) Stochastic transportation-inventory network design problem. Oper Res 53(1):48–60 Snyder LV, Daskin MS (2005) Reliability models for facility location: the expected failure case. Transp Sci 39(3):400–416 Snyder, LV, Daskin, MS, Teo, C-P (2007) The stochastic location model with risk pooling. Eur J Oper Res 179(3):1221–1238 Snyder LV, Scaparra MP, Daskin MS, Church RL (2006) Planning for disruptions in supply chain networks. Tutor Oper Res 2:234–257 Snyder LV, Atan Z, Peng P, Rong Y, Schmitt AJ, Sinsoysal B (2016) OR/MS models for supply chain disruptions: a review. IIE Trans 48(2):89–109

Chapter 18

Process Engineering and Optimization Marcello Braglia, Marco Frosolini, Roberto Gabbrielli and Leonardo Marrazzini

Abstract Process planning plays a fundamental role within modern industrial facilities in that it is devoted to select the manufacturing processes and parameters and the sequences that are needed to convert a part from the initial to the final form. Evaluating the quality of a process plan from both a technological and an economic perspective becomes vital. Indeed, a valid process plan not only ensures the required quality of manufactured parts, but also reduces the production cost. The present chapter is aimed at presenting the essential concepts and the historical evolution of the process planning approaches, starting from a general point of view and then introducing the most important and recent real-world applications. In addition, it introduces the fundamental aspect regarding the interactions between process planning and production scheduling, pinpointing the strengths and weaknesses of the available proposed solutions. Doing so, it addresses students, researchers as well as practitioners.

18.1 Essentials of Process Planning and Project Scheduling (Basic) Process planning (also known as process engineering) is aimed at selecting the manufacturing processes and parameters and the sequences to be used to convert a part from the initial to the final form based on a predefined engineering drawing (Chang and Wysk 1985). This is summarized in Fig. 18.1. It is known that process plans for manufacturing a complex part may have several feasible alternatives. Thus, it is paramount to evaluate the quality of a process plan from both a technological and an economic perspective. A valid plan not only ensures the required quality of manufactured parts, but also reduces the production cost. Noteworthy, process planning activities in the past strongly depended on the knowledge and experience of engineers and technicians. The most important M. Braglia (B) · M. Frosolini · R. Gabbrielli · L. Marrazzini University of Pisa, Pisa, Italy e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_18

399

400

M. Braglia et al.

Fig. 18.1 Process planning

advances have been attained over the years with an ever-increasing adoption of information technology based solutions and with the introduction of the well-known Computer Aided Process Planning (CAPP) systems. One of the most important goals of this development, fostered by an increasing number of research contributions (Yusof and Latif 2014), was the integration of all aspects of manufacturing and, therefore, to develop a Computer Integrated Manufacturing (CIM) environment. In brief, at least in theory, a CIM system should be able to accomplish the full flexible automation of every manufacturing activity and the coordination and optimization of the entire production system (Fig. 18.2). Process planning and, naturally, the corresponding CAPP systems play a key role in the integration process, in that they represent the fundamental link between design and manufacturing (Xu et al. 2011). The process plan starts from the available designs and moves on considering the existing constraints in the field (for instance, resources, machines and materials), allowing the production process to run smoothly. This leads to the definition of process planning as a “systematic definition of both methods and tools by which a product is manufactured efficiently and economically” (Tulkoff 1987).

18 Process Engineering and Optimization

401

Fig. 18.2 CAPP as key component of CIM

Owing to these considerations, a great deal of research has taken place with respect to the integration of other up/downstream systems (such as, for instance, Computer Aided Design—CAD, and Computer Aided Manufacturing—CAM) and CAPP systems (Yusof and Latif 2014). Since many years, this has led to two basic approaches to CAPP that are, namely, variant and generative process planning. A third approach is also available (Hybrid CAPP), which further extends the generative process planning. In the variant approach, specific computer-based applications use tables and lookup procedures to individuate similar components and to build feasible plans. The key principle is that similar parts can be produced with similar plans (Boer et al. 1990). Generally, the human planner classifies the parts and inputs their data, looks for similar items to retrieve a standard plan and finally, moving from this initial draft, creates variants, which may suit the specific requirements of the actual process. This has been for long the most used methodology to computerize the planning process. The variant approach mostly uses the well-known Group Technology part coding, where several attributes and characteristics allow identifying and univocally coding such parts. Those items that share a given number of characteristics are put together to form families. Then, the whole planning is generated for such families and variants are developed starting from the same archetype. The system is characterized by a high maintainability, but strongly depends on the knowledge of the human planner (Koenig 1990). The generative process planning requires a greater computational intelligence, in that the algorithm automatically creates a process plan for each new component (Boer et al. 1990; Chryssolouris 1992). This means that the whole process requires very limited human intervention and only relies on the information that is available in the manufacturing database. In brief, the system retrieves the design model from a CAD file and produces the necessary operations to manufacture the item. Besides, it structures such operations in a technically feasible sequence (Chryssolouris 1992).

402

M. Braglia et al.

The most significant difficulty, in this case, is represented by the necessity of encoding the required knowledge within a computer program, imitating the behavior and the decision making approach of the human planner. Interestingly, other planning functions such as machine selection, tool selection, process optimization can also be automated using generative planning techniques. Compared to the variant approach, the generative process planning systems have significant advantages. They are able to produce feasible and consistent plans rapidly, in the case of both existing components and completely new items. Moreover, they can potentially integrate with the externally available automation systems to gather detailed control information (Chryssolouris 1992). It is noteworthy that part descriptions and technical designs can be input for the CAPP system in different ways, all affecting the resulting degree of automation. With the technological advances in CAD systems, the process that provides the part features to the computer-based planner is no longer an issue and 3D solid models may be used instead of the traditional 2D drawings. These often hold and integrate information on the materials as well, further enriching the available information and allowing the transformation of the CAPP into a true expert system (Shah et al. 1991). Briefly, some of the most important characteristics and desiderata of an effective CAPP can be summarized as follows (Xu et al. 2011): • Ability to interconnect with up and down-stream processes, such as design, scheduling and control; • Modularity and flexibility within a unique, generally valid solution. At present, this represents a significant issue, in that the available process planning systems are very specific with respect to the analytical models and the knowledge bases (Yuen et al. 2003); • Ability to provide effective knowledge acquisition. In the early development stage of such systems, which roughly took place from 1965 to the early 90’s, most of the attention was addressed to the geometric aspects of the problem, with the aim of effectively integrating CAD and CAPP systems. Most of the solutions used—and still use—feature-based technologies. It is important, at this point, to distinguish between manufacturing and design features. Manufacturing features can be described as a physically distinguishable portion of a part, such as a surface, a hole or a chamfer. A design feature is fundamentally the same thing, but it exists to attend a specific purpose in the product, thus having a broader meaning. Planning systems require features to input data. There are two different approaches (Shah et al. 1991): (i) feature recognition, that determines the existence and the characteristics of a feature starting from topology and geometry, and (ii) design by features, where appropriate libraries of predefined features are available to build parts. In this case, the geometry of the single feature is fixed, but the true dimensions are managed as variable parameters. A recent study (Babic et al. 2008) has highlighted the most significant issues of feature recognition. Clearly, the correct extraction of the geometric primitives from a CAD file, the definition of features and the feature pattern recognition have a significant impact on the success of the planning process.

18 Process Engineering and Optimization

403

Since the early 80’s researchers have attempted to include complex inferential reasoning capabilities within CAPPs (Welband 1983). These systems are built upon three main components, i.e., the knowledge base, the inference engine and the user interface. Their most attractive aptitude is that of accumulating knowledge as time passes. Further, they can derive features, rules and knowledge using available examples and the training process and they can successfully face exceptions and irregularities that would otherwise hinder or stop the inferential process. Therefore, the process planning can cope with the dynamic nature of the manufacturing environment. All these approaches could be integrated with a scheduling system to modify the plan at runtime, adaptively and according to the current resource availability. The scheduling of activities represents one of the fundamental components that contribute to the integration of the various phases of manufacturing, together with the CAPP. As stated previously, the purpose of CAPP is to determine and optimize feasible process plans that can be manufactured economically. However, in doing this it generally considers only technological constraints, neglecting the current time and resource availability limitations. Naturally, this deals with scheduling. The scheduling problem is defined as the allocation of operations to machines in time, taking into account currently available resources, and its output is a capacitated sequence of operations (Conway et al. 1967). For instance, a typical Job Shop scheduling problem searches for the sequence of n jobs on m available machines, under predefined technical constraints, that is optimal with respect to some performance criteria (French 1982). Since many years, research on scheduling has been classified following four main parameters (Conway et al. 1967): job arrival patterns, number of available machines, flow patterns and, finally, the criteria with which the scheduling will be evaluated. The literature on scheduling models is at present rather extensive (see, for instance, Garey et al. 1976; Adams et al. 1988; Schutten 1998; Zhang et al. 2007). Among the most important considerations, it emerges that scheduling problems are NP-complete or, in other words, the hardest combinatorial optimization problems (Garey et al. 1976). During the last decades, several optimization and approximation algorithms have been proposed to find optimal solutions. Notably, as stated by Framinan and Ruiz (2010), very little has been written to bring models and procedures into practice, originating the so-called “gap” between the theoretical body and the practice of scheduling (McCarthy and Liu 1993). Furthermore, in most of the traditional approaches, process planning and scheduling are carried out sequentially (Li et al. 2010) and without any real-time feedback, generating a true obstacle to improve the productivity and the responsiveness of manufacturing systems and generating a significant number of issues.

404

M. Braglia et al.

18.2 Interaction of Process Planning and Production Scheduling (Advanced) As stated in the previous paragraph, many traditional CAPP systems do not integrate scheduling while generating the process plan. There are a few exceptions, like the early work of Zijm (1995). In brief, scheduling was generally carried out distinctly, only after the process had been completely defined by the process planner. This impeded to account for optimality (Framinan and Ruiz 2010; Li et al. 2010). Moreover, process planners tended to elaborate their plans on the assumption of ideal conditions, choosing the best machines to get the best process plans, but without considering their current availability (resource competition). This usually leads to queues and excess allocation of some available resources (Phanden et al. 2011). Even worse, it also may result in the necessity of modifying the plan and repeating the complete loop, with a significant loss of time. Indeed, while process planning stresses the technological aspects of the processes, scheduling is devoted at optimizing time, thus resulting in conflicting targets. As an example, the time delay between the two phases may determine the necessity of regenerating the process plan: this is mainly due to the dynamic nature of the real systems, where scheduling issues produce new technical constraints that may require a complete redevelopment. Conversely, it could also depend on the schedule itself. In this case, the unavailability of the required resources determines the unfeasibility of the program and requires, again, a complete revision of the whole plan. Bottlenecks and unavailability are two common outcomes of such two-step processes (Phanden et al. 2011). Furthermore, the optimization of both phases has always been considered a single-criterion process. In the real world, it turns out that the use of multiple optimization criteria might lead to better results (Li et al. 2010). Some recent studies reveal that up to 30% of the jobs within a shop should be readdressed over time to consider sudden operative issues and new technical constraints. This means that those processes are not valid and need to be modified accordingly, before starting the production process. These are some of the major obstacles to the realization of a CIM system and they have been repeatedly pinpointed over time in the last decades. Owing to these considerations, process planning and scheduling should be considered complimentary functions. Over time, researchers have proposed to integrate them to achieve the global optimization of product development and manufacturing. This is extremely important, because in the recent past companies have progressively increased the customization of their products, in a continuous search for competitiveness. Generally, this results in smaller production lots and a growing need of flexibility. Besides, the contemporary advances of information systems have led to a better integration with suppliers and customers within the supply chain. All these aspects force the manufacturers on the one hand to respond faster, timely and accurately to customer needs while, on the other hand, creating more efficient and effective networks with their suppliers. Obviously, the success in fulfilling these goals strictly depends on the capability of reconciling the two aforementioned conflicting aspects.

18 Process Engineering and Optimization

405

Indeed, firms must translate the product specifications into technically feasible process plans as fast as possible, while integrating them into the existing production schedule. The compelling requirement is to succeed in correctly accommodating the current and the upcoming workloads, considering the status of the human resources, machines, equipment and tools, and the availability of materials (Li et al. 2010). Researchers have proposed to support these capabilities integrating CAPP/CIM systems with production scheduling. Thus, concurrent planning and scheduling represents a valuable tool to enhance the ability of manufacturing companies to adapt efficiently to rapidly changing conditions. In particular, it could effectively improve the ability of manufacturing companies to adapt efficiently to changing conditions, and yield significant performance improvements, such as, for example, shorter lead times, increased resource utilization, enhanced due-date performance and coordination between customers and suppliers. One of the major limitations to this integration resides in the fact that while the automatic planning eliminated the necessity of human intervention, solving integrated process-planning and production-scheduling problems still requires a significant level of interaction. The scheduling stage needs to interact with numerous external information sources. To overcome such limitations, researchers are currently trying to implement specific decision-making features to address problems at multiple levels and with different criteria and to include the ability to investigate alternative solutions in the scheduling system. Efforts to develop decision-support frameworks in the areas of planning and scheduling have been reported in Phanden et al. (2011). Technical challenges in supporting integrated process-planning and production-scheduling decisions in a complex and dynamic environment are multiple. The major scheduling challenges include the presence of multiple sources of uncertainty. However, while considerable progress has been made with respect to both the software technologies for process planning and the finite-capacity production scheduling, for many years very little attention has been given to issues of integration. Except for a few attempts, often in the context of small manufacturing environments, processplanning and production-scheduling activities have been handled independently. Process alternatives are compared strictly based on pure technical considerations, and plans are developed without consideration of the current availability of the resources. Similarly, production scheduling is performed under fixed process assumptions and without regarding the opportunities that process alternatives can provide for acceleration of production flows. Even with the support of sophisticated CAPP systems and scheduling techniques, process planning and production scheduling remain highly interactive processes, where the user evaluates alternative decisions based on personal experience and knowledge. Thus, today it is extremely important to address the research efforts towards the Integrated Process Planning and Scheduling (IPPS) systems. These could effectively help reducing the scheduling conflicts, the flow time and the work in process, with a significant beneficial effect on the use of the production resources. The best way to integrate is by increasing information exchange between process planning and scheduling. In recent reviews, the authors underline that several classification

406

M. Braglia et al.

Fig. 18.3 NLPP approach

schemes have been suggested by various researchers (Gindy et al. 1999) and that three main approaches of integration are available: Non-Linear Process Planning and Scheduling approach (NLPP), Closed-Loop Process Planning and Scheduling approach (CLPP) and Distributed Process Planning and Scheduling approach (DPP). NLPP (Fig. 18.3) considers three different types of flexibility to generate multiple process plans for the jobs, before they enter the shop floor. The operation flexibility refers to the possibility of executing an operation on more than a machine. The sequencing flexibility represents the possibility of interchanging the sequence of the manufacturing operations. Finally, the processing flexibility is the possibility of producing the same manufacturing feature with alternative operations. NLPP is mainly focused on static shop floor situations (Gaalman et al. 1999). Multiple process plans are evaluated and ranked according to some decision criteria, such as, for instance, the machining time or the production time. All ranked process plans are stored in a knowledge database and the first ranked plan is sent to the shop floor, as soon as needed. Naturally, if the first priority plan is not suitable with the current state, the other available schedules may be used, following the priority ranking. Recently, researchers (Moon et al. 2002; Zhao et al. 2006) have introduced several improved methodologies, such as Genetic Algorithm (GA)-based, hybrid Genetic Algorithm-Tabu Search (GA-TS) approaches, Particle Swarm Optimization (PSO) or Symbiotic Evolutionary Algorithms (SEA) to simultaneously deal with process planning and job shop scheduling and, finally, to improve Non-Linear IPPSs. They develop efficient genetic representations and consider specific criteria, such as the minimization of the makespan, to find optimal solutions. With respect to multiple criteria optimization, a Multiple Objective Tabu Search (MOTS) algorithm has been successfully proposed (Baykaso˘glu and Özbakır 2009) to increase flexibility while reducing the process plan overall costs. Other authors (Rajkumar et al. 2010) have used a multi-objective Greedy Randomized Adaptive Search Procedure (GRASP) to minimize makespan, maximum workload, total

18 Process Engineering and Optimization

407

Fig. 18.4 CLPP approach

workload, tardiness and total flow time within a flexible job-shop environment. In any case, the NLPP systems suffer from a significant weakness, in that they do not benefit from a true real-time feedback. The CLPP approach (Fig. 18.4) uses the dynamic feedback from production scheduling, with information on the current availability of resources, to generate the process plans. In brief, the scheduling phase communicates the present availability of the machines on the shop floor to the process planning. Every time an operation is completed on the shop floor, a feature-based workload is generated to determine the next operation and to allocate the necessary resources. In general, due to the necessity of real-time bi-directional communications, the process planning and scheduling departments must be completely reorganized (Iwata and Fukuda 1989). As in the case of NLPP, the use of meta-heuristics has recently become rather extensive. Some authors (Sugimura et al. 2003) have proposed an IPPS model in which the planning phase determined the sequence of machining features using a Genetic Algorithm, whereas the optimal sequence of machining equipment was selected using Dynamic Programming (DP). The proposed model can consider the future schedule of the available machines, with the objective to minimize the total machining and set-up time. In Wong et al. (2006), the authors develop an agentsbased multi-stage negotiation protocol scheme for IPPS in a job shop. The model uses part agents and machine agents to represent parts and machines respectively. Part and machine agents negotiate to establish the actual schedule using process plans and operation details from an AND/OR graph. In addition, the negotiation protocol can handle multiple tasks and many-to-many negotiations. The authors conclude that the proposed procedure performs better than meta-heuristics when dealing with local objectives, such as, for example, the optimization of the parts flow time. Hybridbased multi-agent system have been used as well, extending the previous studies. Such hybrid approaches prove to be effective for large scale scheduling problems and provides a better global performance with respect to other available models. In DPP, both process planning and production scheduling are performed simultaneously (Fig. 18.5). The DPP engine separates process planning and production scheduling tasks into three phases, namely the preplanning, the matching and the final planning phase. To begin with, the process planning function analyzes the job based on the product data. The features and feature relationships are recognized and

408

M. Braglia et al.

Fig. 18.5 DPP approach

corresponding manufacturing processes are determined. At this point, the required machine capabilities are punctually estimated. The second phase matches the required job operations with the operation capabilities of the available production resources. The integration occurs at the point when resources are available and the job are required. The result is a true dynamic process planning and production scheduling, effectively constrained by real time events. This model is the only one that integrates the technical and capacity-related planning tasks into a dynamic fabrication planning system (Larsen and Alting 1990). Unfortunately, it requires extremely high capabilities from both hardware and software applications. One of the first DPP systems, proposed by Aanen et al. (1989), consists of a hierarchical approach in which the primary and secondary objectives are respectively to satisfy the due-dates of an order and to minimize the change-over and idle times of the available machines. The planning problem is initially solved and the resulting output becomes the input for the scheduling engine. Feedback information is provided to the planning level as soon as the output of the scheduling is not satisfied. Many interesting works have followed. Recently, Cai et al. (2009) presented an IPPS methodology with a multimachine setup planning approach using a GA. This is specifically addressed to solve the adaptive setup planning (ASP) and has been extended to solve the multi-machines setup-planning problem. The GA-based ASP shows to be able to respond quickly and effectively to changing shop floor situations. In Wang et al. (2010) the authors propose an IPPS based upon a GA-based approach for job shop machining that consists of a generic setup planning and an adaptive setup merging to optimize cost, quality, makespan and machine utilization. Based on the current machines availabilities and capabilities, the proposed approach generates the setup plans adaptively.

18 Process Engineering and Optimization

409

Artificial Neural Networks (ANNs) have proved very effective helping machines to learn. Indeed, in Ueda et al. (2007), the local decisions related to the selection of the part to process and the machine to use are taken by machine agents. Each machine learns the state of the shop floor using a specifically trained ANN. More recently, Li et al. (2010) presented an agent-based approach for IPPS in a job shop problem. This approach contains three agents and the corresponding databases and has the twofold objective of minimizing the production time and the makespan. Job agents and machine agents are used to optimize alternative process plans and schedules. Operational, sequential and processing flexibility are properly considered within the planning process. When changes occur at the shop floor, not allowing the schedule to be carried out, the machine agents negotiate with the other agents to start a rescheduling process. Though many approaches have been recently proposed and verified, some problems still remain unsolved. In particular, it appears evident from the available literature that the world of IPPSs is highly fragmented. Each one of the proposed models shows strengths and weaknesses. Therefore, in the future it will be necessary to integrate the different models of IPPS with the aim of solving larger problem sets. The current fragmentation still impedes the realization of real world software applications that can be used widely in manufacturing organizations.

18.3 Special Topics in Process Engineering (State-of-the-Art) At present, researchers are continuously pursuing new goals and many interesting features are being studied in detail. Recent research on the integration of CAPP and scheduling aims at solving and improving some weak aspects that are still present in the numerous scientific contributions that have dealt with this topic. In particular, new approaches about how to include the planning of the setup operations and their times are proposed in order to improve the coordination between the process planning and scheduling (Haddadzade et al. 2016). The current approaches are mainly focused on deterministic constraints of jobs, even if the reality of scheduling is characterized by many inefficiencies, such as machine and tool failures, or workpiece deformation, that impact both the scheduled processing times and the availability of the machines. The presence of uncertainty based on processing time scenarios is taken into consideration by Jin et al. (2016), while new dynamic ways to improve the iterative integration between the process planning and the current status of the shop floor have been proposed with different optimization approaches (Yu et al. 2015). The importance of the energy consumption is growing steadily and its minimization in the integration between process planning and scheduling has been recently used as on optimization criterion (Zhang et al. 2016).

410

M. Braglia et al.

Nowadays, the operative production environment is strongly dynamic due to many unplanned events, such as machine malfunctions, poor quality or lack of material and resources, product personalization, dynamic product demands and shortened product lifecycle. In a fluctuating shop floor environment, a process plan that is generated in advance may turn out to be unsuitable or unusable to the targeted resources. In this production context, it is necessary to have a flexible and dynamic process planning and control system that is able to adapt in response to changing manufacturing resource availability, production uncertainty, and dynamic machining conditions. Many recent literature contributions are proposing specific configurations for the integration between production planning and shop-floor scheduling to obtain an adaptive system that can adjust the production plan in real time in accordance with unplanned and sudden modifications. A severe criticality of the integration of CAPP and scheduling appears when the manufacturing resources are networked and geographically distributed. Recent progress in information technology and the pervasive application of web services largely influence the manufacturing systems and, in particular, the networkdistributed manufacturing environment. Digitalization and networking of enterprises implies the challenges of the accessibility and interoperability of the information in a distributed context. In this manufacturing configuration, resources that are in different factories and that are characterized by different operative features, such as productivity and processing times, complete the process. The networked manufacturing offers several advantages in the current competitive market, by shortening manufacturing cycle time, maintaining the production flexibility and, thereby, achieving several feasible process plans. To improve the integration between the process planning and the scheduling in this specific manufacturing context, several approaches have been proposed and discussed, using different optimizations approaches, such as genetic algorithms and mobile-agent based approach (Manupati et al. 2016). The problem of managing the manufacturing in network production systems also concerns the stand-alone CAPP. In particular, it interests the definition of the process planning of items that are produced by different equipment that are located in a network of geographically distributed shop floors and that can communicate using the web. In this context, some architectures of web-based structures for the sharing of the information between the design phase via CAD systems and the process definitions have been proposed (Varela et al. 2016). The aim of these systems is to ensure the correct flow of information between different design and manufacturing sites for the definition of the right process plan. In a similar context of web-connected multiple factories, a recent research activity concerns cloud manufacturing (see also Chap. 12) and automatic process planning. Higher quality products at low cost, with quick delivery and shorter times between successive product generations are required by the market. The cooperation between different companies is becoming strong, product-dependent, customer-oriented and dynamic. Manufacturing jobs are diversified and urgent. The shop-floor environment is becoming more and more geographically distributed across corporate and national boundaries. Flexibility and adaptive capacity to unplanned deviations on shop floors where machining processes should be modified dynamically are required.

18 Process Engineering and Optimization

411

This leads to web-based service-oriented cloud manufacturing, that promises to overcome the current limitations of rigid systems, standalone software, centralized resource utilization, unidirectional information flows and off-line decision-making. CAPPs should be responsive and adaptive to the changes of production capacity and functionality. Cloud manufacturing provides cost-effective, flexible and scalable solutions to companies by sharing manufacturing resources as services. The idea of cloud manufacturing is clearly based on cloud computing and is a new-generation service-oriented approach to support multiple companies in deploying and managing services for manufacturing operations over the Internet. Process planning must be able to accommodate the change and distribution of manufacturing resources and materials processing tasks, in collaboration with different process planners. In other words, it should support collaborative process planning among the planners at different places, and improve instantaneous communication among each other. The current proposed cloud structures for CAPP (Mourtzis et al. 2016) concerns internet- and web-based service-oriented systems for machine availability monitoring and process planning. These collect and share the information about the shop floor for adjusting the process planning using Cloud Platforms. Based on the real-time information on machines and their availability, it is possible to create process plans that are adaptable to changes through well-informed decision-making. Another research area currently studied in both Industry and Academia is the integration of additive and subtracting manufacturing (hybrid manufacturing) and their combined process planning. The main peculiarity of additive manufacturing is the ‘layer by layer’ additive construction manner, which makes it possible to manufacture parts with extremely complex geometries without the use of fixtures, tooling, mold or any other additional auxiliary (see also Chap. 23). In conventional subtractive manufacturing, the starting point of the manufacturing is the raw block and the unwanted material is removed during the production. Additive manufacturing allows the production of very complex geometries, but unfortunately the current precision and surface quality are still poor. The synergic combination with conventional subtractive machining, such as milling and grinding, is used to produce fine surfaces after an additive process and, consequently, to obtain the precision that is required by many industrial sectors, such as automotive or aeronautical (Newman et al. 2015). The adoption of subtractive machining can be considered a remanufacturing process. In this case, there are various options to recover/complete unfinished or missing shapes. It is possible to manufacture the whole item again, to add material to the missing area and to fill again material after cutting a part, to get a surface for the additive process. Another possible combination of subtractive and additive manufacturing concerns processes such as the bead-based deposition (i.e., laser cladding, metal inert gas welding, thermal spray), that can be employed for component coating, repair or building features upon an existing component that has been processed by a conventional machine. Hybrid manufacturing has the ability both to consolidate the advantages of independent processes and to mitigate the disadvantages. These kinds of hybrid manufacturing processes are characterized by very peculiar features that are not yet managed effectively by the modern CAPP systems and that make the capabilities of the whole

412

M. Braglia et al.

manufacturing system underutilized. Hybrid manufacturing requires new process planning methods based on the adaptive combination of process capabilities and utilization of the manufacturing equipment. Newman et al. (2015) propose a framework that combines additive, subtractive and inspection processes on a single platform, making it possible to decide how to reuse or to remanufacture existing unfinished items based on their dimensional information thanks to inspection techniques such as touch trigger probing. This approach enables the item to be further manufactured by additive/subtracting processes providing new enhanced functionalities. Using the product information directly obtained from the CAD file and the existing geometry, identified with an inspection device, the process-planning framework provides a set of feasible manufacturing options. This selection depends on process capabilities, process planning knowledge, geometry constraints and manufacturing knowledge. To improve the usability and to reduce the inconsistencies that often are present in traditional CAPP systems, virtual reality and haptic virtual machining applications have been recently proposed as a valuable added tool (Fletcher et al. 2013). Within these models, an operator can simulate the machining of a simple part using virtual manufacturing processes via a haptic routing interface. The virtual system is more intuitive and requires less mental workload than traditional manual methods, particularly for novice process planners. A research topic that many academic and industrial researchers are still investigating is the definition of the CAPP features that are specifically required for specific production contexts. Indeed, the characteristics of the items that must be produced, such as shape, tolerances, finishing of the surfaces, along with those of the machines and of the production processes, require CAPPs that are specifically conceived and structured. Currently available CAPPs are not general purpose and effective for all applications. Fountas et al. (2014) propose a group of manufacturing functions, such as the selection of raw materials, cutting tools, part support and positioning, for the design via CAD of aerospace structural parts where the products are unique, dissimilar, and one-of-a-kind. Jong et al. (2015) have developed an automated CAPP that integrates CAD and CAM for the machining process of plastic injection mold components, of which the machining process is complex and continuously changing. Using feature recognition, the part design geometry features of CAD are converted into manufacturing features and then compared with the identified manufacturing features based on group technology conditions in order to automatically generate the manufacturing program. In conclusion, each of the research topics described above confirms the importance of the subject, its fragmentation and the numerous issues and weaknesses that still must be addressed. However, due to its great potential, it is important to investigate how to realize CAPP systems that are suitable for use in modern geographicallydistributed highly-digitalized factories.

18 Process Engineering and Optimization

413

18.4 Further Reading An early review on process planning from an historical perspective is provided by Tulkoff (1987). Job shop scheduling has been one of the most extensively studied problems in the production management literature, starting with the book by Convay et al. (1967), see e.g. Blazewicz et al. (1996) or Brucker et al. (1994). Adams et al. (1988) introduced the shifting bottleneck procedure for job shop scheduling which is intuitively very appealing and since its publication raised considerable research interest, see Schutten (1998), who shows how to adapt the procedure to a variety of practical settings. A more detailed explanation of the Shifting Bottleneck procedure is presented in Chap. 19 of this volume. Zhang and Merchant (1993) and Huang et al. (1995) discuss approaches to the integration of process planning and job shop scheduling, see also Zijm (1995). An algorithm to deal with distributed scheduling problems in e.g. a cloud manufacturing environment is present by Chan et al. (2005). Cloud manufacturing itself is discussed in Chap. 12.

References Aanen E, Gaalman GJ, Nawijn WM (1989) Planning and scheduling in an FMS. Eng Costs Prod Econ 17(1–4):89–97 Adams J, Balas E, Zawack D (1988) The shifting bottleneck procedure for job shop scheduling. Manage Sci 34:391–401 Babic B, Nesic N, Miljkovic Z (2008) A review of automated feature recognition with rule-based pattern recognition. Comput Ind 59(4):321–337 Baykaso˘glu A, Özbakır L (2009) A grammatical optimization approach for integrated process planning and scheduling. J Intell Manuf 20(2):211–221 Blazewicz J, Domschke W, Pesch E (1996) The job shop scheduling problem: conventional and new solution techniques. Eur J Oper Res 93:1–33 Boer CR, Petitti M, Lombardi F, Simon JP (1990) A CAPP/CAM expert system for a high productivity, high flexibility CNC turning center. CIRP Ann Manuf Technol 39(1):481–487 Brucker P, Jurisch B, Sievers B (1994) A branch and bound algorithm for job-shop scheduling problem. Discret Appl Math 49:105–127 Cai N, Wang L, Feng HY (2009) GA-based adaptive setup planning toward process planning and scheduling integration. Int J Prod Res 47(10):2745–2766 Chan F, Chung S, Chan L (2005) An adaptive genetic algorithm with dominated genes for distributed scheduling problems. Expert Syst Appl 29:364–371 Chang TC, Wysk RA (1985) An introduction to automated process planning systems. Prentice Hall, Englewood Cliffs, New Jersey Chryssolouris G (1992) Manufacturing systems: theory and practice. Springer, Heidelberg Convay R, Maxwell W, Miller L (1967) Theory of scheduling. Addison-Wesley Publishing Company, Reading, Massachusetts, USA Fletcher CA, Ritchie J, Lim T, Sung R (2013) The development of an integrated haptic VR machining environment for the automatic generation of process plans. Comput Ind 64:1045–1060 Fountas NA, Stergiou CI, Vaxevanidis NM (2014) Programmable support functions to assist process planning for aeronautical parts manufacturing. In: Proceedings of the ASME 2014 12th biennial conference on engineering systems design and analysis ESDA2014, June 25–27, 2014, Copenhagen, Denmark

414

M. Braglia et al.

Framinan JM, Ruiz R (2010) Architecture of manufacturing scheduling systems: literature review and an integrated proposal. Eur J Oper Res 205:237–246 French S (1982) Sequencing and scheduling. Ellis Horwood Ltd, Chichester, UK Gaalman GJC, Slomp J, Suresh NC (1999) Towards an integration of process planning and production planning and control for flexible manufacturing systems. Int J Flex Manuf Syst 11:5–17 Garey E, Johnson D, Sethi R (1976) The complexity of flow shop and job shop scheduling. Math Oper Res 1:117–129 Gindy N, Saad S, Yue Y (1999) Manufacturing responsiveness through integrated process planning and scheduling. Int J Prod Res 37(11):2399–2418 Haddadzade M, Razfar MR, Zarandi MHF (2016) Multipart setup planning through integration of process planning and scheduling. Proc Inst Mech Eng Part B J Eng Manuf 230(6):1097–1113 Huang SS, Zhang HC, Smith ML (1995) A progressive approach for the integration of process planning and scheduling. IIE Trans 27:456–464 Iwata K, Fukuda Y (1989) A new proposal of dynamic process planning in machine shop. In: Proceedings of CIRP international workshop on computer aided process planning, Hannover University, Germany, 21–22 September Jin L, Zhang C, Shao X, Tian G (2016) Mathematical modeling and a memetic algorithm for the integration of process planning and scheduling considering uncertain processing times. Proc Inst Mech Eng Part B J Eng Manuf 230:1272–1273 Jong W-R, Lai P-J, Chen Y-W, Ting Y-H (2015) Automatic process planning of mold components with integration of feature recognition and group technology. Int J Adv Manuf Technol 78:807–824 Koenig DT (1990) Computer-integrated manufacturing: theory and practice. Taylor and Francis, Boca Raton Larsen NE, Alting L (1990) Simulations engineering within process and production planning. In: Pacific conference on manufacturing, Australia, 17–21 December Li X, Gao L, Zhang C, Shao X (2010) A review on integrated process planning and scheduling. Int J Manuf Res 5(2):161–180 MacCarthy BL, Liu J (1993) Addressing the gap in scheduling research: a review of optimization and heuristic methods in production scheduling. Int J Prod Res 31(1):59–79 Manupati VK, Putnik GD, Tiwari MK, Ávila P, Cruz-Cunha MM (2016) Integration of process planning and scheduling using mobile-agent based approach in a networked manufacturing environment. Comput Ind Eng 94:63–73 Moon C, Kim J, Hur S (2002) Integrated process planning and scheduling with minimizing total tardiness in multi-plants supply chain. Comput Ind Eng 43:331–349 Mourtzis D, Vlachou E, Xanthopoulos N, Givehchi M, Wang L (2016) Cloud-based adaptive process planning considering availability and capabilities of machine tools. J Manuf Syst 39:1–8 Newman ST, Zhu Z, Dhokia V, Shokrani A (2015) Process planning for additive and subtractive manufacturing technologies. CIRP Ann Manuf Technol 64:467–470 Phanden RK, Jain A, Verma R (2011) Integration of process planning and scheduling: a state-ofthe-art review. Int J Comput Integr Manuf 24(6):517–534 Rajkumar M, Asokan P, Page T, Arunachalam S (2010) A GRASP algorithm for the integration of process planning and scheduling in a flexible job-shop. Int J Manuf Res 5(2):230–251 Schutten JMJ (1998) Practical Job Shop Scheduling. Ann Oper Res 83:161–177 Shah J, Sreevalsan P, Mathew A (1991) Survey of CAD/feature-based process planning and NC programming techniques. Comput Aided Eng J 8(1):25–33 Sugimura N, Shrestha R, Inoue J (2003) Integrated process planning and scheduling in holonic manufacturing systems—optimization based on shop time and machining cost. In: Proceedings of the 5th IEEE international symposium on assembly and task planning, 10–11 July, Besanqon, France, pp 36–41 Tulkoff J (1987) Process planning: a historical review and future prospects. In: Proceedings of 19th CIRP international seminar on manufacturing systems, major theme: computer aided process planning, Pennsylvania State, USA

18 Process Engineering and Optimization

415

Ueda K, Fuji N, Inoue R (2007) An emergent synthesis approach to simultaneous process planning and scheduling. Ann CIRP 56(1):463–466 Varela MLR, Araýjo AF, Putnik GD, Manupati VK, Anirudh KV (2016) Web-based decision system for effective process planning in network manufacturing environment. In: 11th Iberian conference on information systems and technologies (CISTI), 15–18 June 2016 Wang L, Cai N, Feng HY, Ma J (2010) ASP: an adaptive setup planning approach for dynamic machine assignments. IEEE Trans Autom Sci Eng 7(1):2–14 Welband M (1983) A review of knowledge acquisition techniques for expert systems. British Telecommunications Research Laboratories Technical Report, Ipswich Wong TN, Leung CW, Mak KL, Fung RYK (2006) An agent-based negotiation approach to integrate process planning and scheduling. Int J Prod Res 44(7):1331–1351 Xu X, Wang L, Newman ST (2011) Computer-aided process planning—a critical review of recent development and future trends. Int J Comput Integr Manuf 24(1):1–31 Yu M, Zhang Y, Chen K, Zhang D (2015) Integration of process planning and scheduling using a hybrid GA/PSO algorithm. Int J Adv Manuf Technol 78:583–592 Yuen CF, Wong SY, Venuvinod PK (2003) Development of a generic computer-aided process planning support system. J Mater Process Technol 139(1–3 SPEC):394–401 Yusof Y, Latif K (2014) Survey on computer-aided process planning. The Int J Adv Manuf Technol 75(1):77–89 Zhang C, Li P, Guan Z, Rao Y (2007) A tabu search algorithm with a new neighborhood structure for the job shop scheduling problem. Comput Oper Res 34:3229–3242 Zhang HC, Merchant ME (1993) IPPM—a prototype to integrate process planning and job shop scheduling functions. CIRP Ann Manuf Technol 42(1):513–518 Zhang Z, Peng RTT, Tao L, Jia S (2016) A method for minimizing the energy consumption of machining system integration of process planning and scheduling. J Clean Prod 137:1647–1662 Zhao F, Zhu A, Ren Z, Yang Y (2006) Integration of process planning and production scheduling based on a hybrid PSO and SA algorithm. In: Proceedings of IEEE international conference on mechatronics and automation, 25–28 June, Luoyang, China Zijm WHM (1995) The integration of process planning and shop floor scheduling in small batch part manufacturing. Ann CIRP 44(1):429–432

Chapter 19

Advanced Production Planning and Scheduling Systems Henk Zijm and Marco Schutten

Abstract In this chapter, we present algorithms for a number of functions of the production planning framework presented in Chap. 5. We focus on models for integrated Capacity and Master Production Planning, Job Planning and Resource Group Loading, and Shop Floor Scheduling and Control. At the Master Production Planning level, we exploit a simple Linear Programming formulation to set appropriate capacity levels and in particular to decide whether a temporary expansion of capacity is needed (e.g., through overtime work). With the same formulation, we decide what end-items are to be produced in which period. By applying the lead time offset procedure that is the heart of Materials Requirements Planning, and using the Bill of Materials information, the same is done on the level of part manufacturing (basic level). Essential in the above procedure are two parameters, the effective overall capacity of each manufacturing shop and the final assembly department, often indicated as the maximum throughput, and the lead times needed to complete a part or product in each department. A significant portion of these lead times may in fact be waiting times in front of individual workstations that are busy. To minimize these waiting times, workload control norms are often used which in turn may influence the effective capacity. An essential question then is what these workloads should be in order to match a desired throughput and production lead time. That question is answered by exploiting a Closed Queueing Network approach that explicitly determines the relation between a preset work-in-process level, throughput and the resulting lead times (advanced level). Finally, we exploit a detailed shop floor scheduling procedure, called the Shifting Bottleneck approach, that basically serves to ascertain that internal due-dates, following from the above defined internal manufacturing lead times are indeed met (state-of-the-art).

H. Zijm (B) · M. Schutten Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, Overijssel, The Netherlands e-mail: [email protected] M. Schutten e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_19

417

418

H. Zijm and M. Schutten

19.1 Introduction: Setting the Stage (Basic) In this section, we consider a generic manufacturing facility, consisting of a parts manufacturing shop followed by an assembly department. Typically, the number of parts produced is limited while a variety of end products can be assembled from specific parts configurations as specified by the Bill of Materials (BoM) of the final product. We assume that purchased items and raw materials needed for parts production can be stocked and that a small inventory of finished parts exists, from which they are picked to enter a final assembly stage. Depending on specific product-market combinations, final products are either produced on customer order or distributed to sales outlets to meet future market demand. Note that, if final assembly is based upon confirmed customer orders, finished product stocks will in principle not exist (apart from a possible small delay in shipping them to the customer); in that case we are dealing with a make-to-stock, assemble-to-order system. A Customer Order Decoupling Point (CODP) is thus located at the finished parts storage facility. If some parts are already customer specific, the intermediate stock point for these parts is also removed, and the CODP is shifted further upstream. However, in order to describe the most generic situation, we in principle accept stock after any possible stage, while in addition Work-in-Process (WIP) inventories are obviously present on the manufacturing shop and assembly floor (cf. Fig. 19.1). Parts processed in the parts manufacturing shops need a certain lead time from picking the necessary materials or purchased items to be used, until their completion and placement in the appropriate finished part stock point (each part type has its own stock location). Before beginning the assembly of a final product, the parts specified by the Bill of Materials are collected from the intermediate storage and subsequently assembled, which again requires a certain lead time. As we will see later, the length of the lead time strongly depends on the workload of the shop; a shop facing a high workload will generally reveal longer lead times, because parts to be processed on a highly utilized machine may experience longer wait times. The impact of lead times on the overall workload is discussed in Sect. 19.4.

Fig. 19.1 Two stage manufacturing facility

19 Advanced Production Planning and Scheduling Systems

419

19.2 Case Study: Injection Molding Machinery (IMM) Injection Molding Machinery (IMM) in the Netherlands is part of one of Europe’s larger manufacturers of capital goods and industrial equipment. IMM was founded in 1968 and quickly specialized in the manufacturing of innovative injection molding machines. A typical machine costs between 0.5 million and 2 million Euro. The world market share is about 4%. Some 80% of all machines are exported to Germany. Most of the about 1300 components required in the assembly of the molding machines are purchased. Only 50 components—the more voluminous part types—are processed in the firm’s own manufacturing department. All purchasing and manufacturing activities are customer order driven. When the production and selling of a new range of injection molding machines began, and a significant increase in demand was expected, the firm decided to install a Flexible Manufacturing Cell (FMC) in their manufacturing department as a replacement for two horizontal milling machines. The cell consisted of a machining center with a 110-slot capacity tool magazine, a parts pallet storage with a capacity of 6 FMC pallets, a rail-guided pallet transport vehicle, and a clamping/unclamping station. Both the exchange of cutting tools in the tool magazine and the clamping and unclamping of parts on the pallets were done manually. A hoist is installed for the transportation of the voluminous and heavy parts to and from the clamping/unclamping station. The parts assigned to the FMC often need some customer-specific processing. Therefore, the number of different NC programs increases steadily. To date, there are over 1000 NC programs for about twenty part type families, which need over 200 different cutting tools. NC programs are written in the Process Planning department. The Manufacturing Planning and Control system of the firm consists of three hierarchical levels: a Master Planning Level, a Production Management Level and a Shop Floor Control Level. The Master Planning Level is basically responsible for generating orders for the FMC. On a strategic level, it is decided that 2–4 injection molding machines per week can be assembled and shipped to the customer. Each injection molding machine has customer-specific parts (mostly variants of specific part types). Engineering makes drawing for these parts. Process Planning subsequently determines the routings of the parts through the manufacturing department and estimates the required processing times. After this, the BoM of a particular injection molding machine is known, and the information will be processes by a Manufacturing Resource Planning (MRP) system. Purchasing of components and raw materials, parts manufacturing and machine assembly are all customer-order driven. The estimated planned lead times for assembly, manufacturing and purchasing are 5, 2 and 6 weeks respectively. Manufacturing is planned such that all the components

420

H. Zijm and M. Schutten

for a particular machine are available and can be kitted two weeks before the assembly of the particular injection molding machines is to occur. These two weeks are the safety lead time and are built in because of problems with the delivery of raw materials. The second level of the production control system is the Production Management Level. At this level, an important objective is to make sure that the planned lead times are met in an efficient way. Therefore, the production management level must tune the manufacturing capacity into the order flow. The Production Management Level receives weekly information from the MRP system about the capacity needed for each machine, including the FMC, covering a period of eight weeks. Based on this information, capacity decisions are made regarding overtime, weekend work, and/or subcontracting. The third level of the production control system is the Shop Floor Control Level. Input to this level is a daily MRP list of orders. During the night, the MRP list is automatically updated. Orders are sequenced according to a critical ratio rule that considers internal due dates (that is, the day manufacturing is scheduled to be finished) and the cumulative processing time required for an order in the manufacturing department. Operators must process the orders as much as possible according to the given sequence. The order on the MRP list of the FMC represents a cumulative workload of one, two or sometimes even three weeks of the machining center. The company faces various problems. There is a pressure to decrease the total lead time, starting with a reduction of the safety time from two to one week. In addition, it lacks an adequate tool to relate the workload to the realized internal manufacturing shop lead times, while at the shop floor level, a more sophisticated sequencing rule is needed. Currently too many deviations of the prescribed sequence occur due to unavailability of raw materials, part programming that is not completed in time, or unavailability of fixtures and cutting tools (that are still in use by other jobs). The company therefore is in need of a sound planning and scheduling mechanism.

19.3 Integrated Capacity and Master Production Planning (Basic) In this section, we present a model for the formulation of an integrated capacity and master production planning problem for a finite planning horizon of T periods (say T weeks). We start with a single stage manufacturing shop and later extend the model to a two-stage system, consisting of a parts manufacturing department followed by an assembly shop. We assume that periodic demand can be reliably forecasted for the full planning horizon. Since capacity is generally limited and in general not sufficient

19 Advanced Production Planning and Scheduling Systems

421

to meet peak demands in case of a volatile market or e.g. in case of strong seasonality patterns, it may be needed to start production of some products early and to store them temporarily, to make sure that eventually all demand can be met. In such as case, inventories basically serve as shifted (allocated) capacities, as opposed to cases where they serve to anticipate normal demand fluctuations (for which safety stocks are generally held). We start with some notations. Let t denote the time period index, t  1, . . . , T . We consider a range of N products, indexed by j, and a manufacturing shop consisting of M workstations, indexed by m. By pjm we denote the processing time, in hours, of one unit of product j on workstation m. The capacity of workstation m in period t is limited by Bmt hours. With hj we denote the inventory costs of one unit of product j per period, while cjm are the cost of processing one unit of product j on machine m. Inventories are calculated at the end of each period. The total lead time of an order of products j through the manufacturing shop equals Lj periods. If an order is placed, it is released at the beginning of a period, say t, and upon completion is added to the stock of product j at the beginning of period t + Lj . To describe the routing of jobs through the system, index ajtms  1 if order j released in period t is processed on workstation m in period s(s ≥ t), while ajtms  0 otherwise. Finally, let the decision variables Qjt denote the size of an order for products j to be released at the beginning of period t, let Ijt be the physical inventory of final products j at the end of period t, and let Djt be the demand for product j which occurs during period t. The following Linear Programming (LP) formulation seeks to minimize the overall production and inventory costs while predicted demand should be met. Since capacity constraints cannot be violated this may cause production occasionally to start relatively early in order to build up inventory such that also peak demands are met. ⎡ ⎤ −Lj  N  N T M T Lj    1−α Qjt + hj Ijt + cjm Qjt ⎦ min⎣ hj 1−α m1 j1 t1 j1 t1 subject to Ijt  Ij,t−1 + Qj,t−Lj − Djt ,

j  1, . . . , N ; t  Lj + 1, . . . , T ;

Ijt  I¯jt ,

j  1, . . . , N ; t  1, . . . , Lj ;

N

t

Qjs pjm ajsmt ≤ Bmt

m  1, . . . , M ; t  1, . . . , T ;

j1 st−Lj +1

Ijt ≥ sjt

j  1, . . . , N ; t  1, . . . , T ;

Qjt ≥ 0

j  1, . . . , N ; t  1, . . . , T .

Note that orders for product j released in period t are driven by the demand forecast of period t + Lj period and update inventory only then. With a time horizon of T periods, period T − Lj therefore offers the last opportunity to release an order for product j. Also, since production of a batch Qjt of product j takes a lead time Lj ,

422

H. Zijm and M. Schutten

it contributes to the work-in-process materials at the manufacturing floor during Lj subsequent periods, for which inventory costs are incurred. We value the batch Qjt at a fraction α k−1 (0 < α < 1) of its final value if it still has k − 1 periods to go till completion (k  1, . . . , Lj ) and hence update the inventory costs accordingly. The contribution of order Qjt to the work-in-process related inventory costs in the objective function therefore equals

1 − α Lj Qjt hj α Lj −1 + α Lj −2 + · · · + α + 1 Qjt  hj 1−α which comes next to the periodic inventory costs associated with stored items. The last term in the objective function reflects the production costs. In fact, one can argue that the latter costs are constant since all demand should be produced anyhow and the costs are time independent, but later we will encounter time-dependent production costs and to ease the presentation we therefore add them already here. If demand is highly volatile or follows a strong seasonality pattern it may be that, despite the possibility to work in advance and store finished products temporarily, the available capacity is still insufficient to meet high demands in some periods. In such a case, companies seek to outsource part of their production or may decide to work in overtime (e.g. a temporary night shift). Obviously, work in overtime is more expensive than work in regular time but on the other hand one may save inventory costs. When allowing for overtime work, one needs to decide which workstation operations (and possibly how much of each operation) are processed in regular time and in overtime. To that end, we define two nonnegative decision variables Qjtms and ∗ that denote the part of order Qjt that is produced on machine m in time period s Qjtms ∗ denote the maximum in regular time and overtime, respectively. Let furthermore Bmt ∗ are the costs of available overtime capacity of machine m in period t, while cjm processing one unit of product j at machine m in overtime. Production limitations are then defined by the following two capacity constraints: N t  

Qjsmt pjm ≤ Bmt m  1, . . . , M ; t  1, . . . , T

j1 st−Lj +1 N t  

∗ ∗ Qjsmt pjm ≤ Bmt m  1, . . . , M ; t  1, . . . , T

j1 st−Lj +1

while furthermore ∗ Qjtms + Qjtms  Qjt ajtms j  1, . . . , N ; t  1, . . . , T ; m  1, . . . , M ; s  t, . . . , t + Lj − 1; ∗ Qjtms , Qjtms ≥0

j  1, . . . , N ; t  1, . . . , T ; m  1, . . . , M ; s  t, . . . , t + Lj − 1.

19 Advanced Production Planning and Scheduling Systems

423

The objective function now becomes ⎡

min⎣

N  T  j1 t1

⎧ ⎫⎤ −Lj ⎨ j −1 N T M t+L ⎬    1 − α Lj ∗ ∗ Qjt + hj Ijt + (cjm Qjtms + cjm Qjtms ) ⎦ hj ⎩ 1−α ⎭ j1 t1

m1 st

In comparison with the initial formulation, the last summation in the objective function now takes into account in which period each operation takes place. The reason is that we may decide per workstation, and hence per period during the production lead time, how much will be produced in regular and how much in overtime. Naturally, regular production is cheaper, hence we attempt to produce as much as possible in regular time. If however, inventory-holding costs are relatively high, it may be advantageous to use the overtime option for production short in advance of needs, instead of producing earlier and pay more holding costs. The latter option is also more realistic if it is difficult to forecast periodic demand too long in advance. The capacity and production planning problem can also be extended in another way. Consider the situation introduced in Sect. 21.1 (see Fig. 21.1) in which a final assembly department is preceded by a parts manufacturing shop. Then a logical procedure is to first plan the assembly department, using the Linear Programming formulation, followed by the planning of the parts manufacturing department. The ¯ pt follows from the just defined production plan of the periodic demand for parts D final assembly department, as follows. Let the parts be indexed by p (p  1, . . . , P), and let bpj denote the number of parts of type p that are assembled in one product j. ¯ pt for parts p is defined by Then the demand D ¯ pt  D

N 

bpj Qjt , p  1, . . . , P; t  1, 2, . . .

j1

which in turn drives the parts manufacturing plan, using the off-set lead times L¯ p of parts p(p  1, . . . , P). When linking the two models but solving them separately, the overall solution may still not be optimal in terms of overall costs. When formulating the integrated parts manufacturing—final assembly model, one immediately sees that the integrated objective function is the sum of the two objective functions in the separated (although linked) models. Since also the two constraint sets of the separated problems are not interfering, the overall minimum of the integrated problem is always smaller than the sum of the minima of the separated problems. Several alternative formulations have been proposed by various authors, although the formulation above with the integration of product- and part-dependent lead times has to the best of our knowledge not been proposed earlier. Hopp and Spearman (2008) discuss an LP-formulation for a single-stage multi-product problem in which all orders are completely produced in the period in which they are released. They also allow for backorders, discuss overtime production and present a model for workforce planning in which costs are incurred with any change of staff level (hiring or firing), in this way forcing the management to keep the workforce as stable as possible.

424

H. Zijm and M. Schutten

Another issue is whether reliable forecasts on a product level be realistically made over a longer planning horizon. To that end, Hierarchical Production Planning (see also Chap. 12) offers an alternative by first planning capacity on an aggregate (product family) level in which a family consists of products with approximately the same capacity requirements, see Hax and Candea (1984), while later disaggregating this plan to production plans on an individual product level. Bitran et al. (1982) apply this approach for a two-stage manufacturing system similar to ours but focus on a single resource (manual labor) only while all lead times are equal. Another way to reduce the need to generate demand forecasts that may turn out to be less reliable over a longer time period is to integrate the linear programming models in a rolling horizon framework in which forecasts are periodically updated. After solving the LP (or LP’s in a 2-stage problem) only the first decisions are implemented, say to period T˜ < T − maxj Lj − maxp L¯ p after which the time index is shifted (period T˜ + 1 becomes period 1), and the whole procedure is repeated. Obviously, capacity profiles have to be updated carefully as well, taking into account the effects of decisions that have been taken in or shortly before period T˜ since they are affecting the free capacity of workstations also after (as a result of longer lead times) as well as the initial inventory positions but such procedures will not be discussed in detail here. Once more, we wish to stipulate that the linear programming model basically serves to smooth capacity over time, making sure that expected demand will be met by balancing the capacities against inventories, possibly with additional options such as overtime work or outsourcing (the latter not discussed here in detail). The main idea is to cope with highly volatile, but predictable, demand patterns, to make sure that parts and products are ready when needed. They are by no means meant to deal with short term demand fluctuations for which stochastic models are more appropriate, both to model workstations and order routings (see the next section) as well as to determine safety stock levels (Chap. 20 of this volume). However, it is important that the lead times Lj used in this section realistically reflect the actual situation. It is well known that the off-set lead times used in e.g. MRP systems are basically larger than necessary, to cover all possible foreseen and unforeseen effects (batching, machine failures, quality problems). Besides, a high workload typically causes lead times to grow quite fast, leading to expediting work orders at the cost of others, etc. For that reason, workload control is often applied to keep lead times stable. How workload control is affecting the effective capacity (throughput) and the resulting lead times of a manufacturing shop is the topic of the next section.

19.4 Throughput and Lead Time Under Workload Control: A Queuing Network Analysis (Advanced) To determine how the average lead time of a part of type k in the manufacturing shop is impacted by the load of the shop, we model the parts manufacturing department as a Closed Queueing Network (CQN) with multiple part types, see e.g. Chen and Yao

19 Advanced Production Planning and Scheduling Systems

425

Fig. 19.2 Manufacturing shop with single machine workstations and a fixed number of jobs (parts)

(2001). The processing time of a part at workstation m is exponentially distributed with rate μm (m  1, . . . , M ). At each workstation, parts are processed according to a FCFS (first-come, first-served) routine. Each part needs a card attached, at Station 0, to guide it through the system. Upon completion of its final operation, the job is returned to Station 0 where the card is detached, to be used for the next job. Because we are interested in the maximum production capacity of a system with say N cards, we assume that there are always jobs present to be loaded if a card is freed. Station 0 is called the load/unload station; we assume that loading and unloading time is negligible. Define routing probabilities as follows: P0m the probability that the first operation takes place at station m, m  0, 1, . . . , M . Pkm the probability that a job, after visiting station k, next visits station m, for k, m  0, 1, . . . , M . Pm0 the probability that a job, after being processed at station m(m  0, 1, . . . , M ), leaves the system. The fact that both processing times and routings are probabilistic reflects a part type aggregation step; in fact, we work with a generic part, representing various underlying part types that may differ in processing time and routing. All assumptions can be relaxed, the current simple model is used as a starting point to present an important analysis method, called Mean Value Analysis (MVA), see Reiser and Lavenberg (1980). Clearly, the N cards are a means to implement a workload control rule, i.e., to allow a maximum of N jobs to be present simultaneously in the workshop. To determine the maximum system throughput (capacity), we therefore have to analyze a system that is always fully occupied, i.e., one in which all N cards are continuously circulating. For ease of presentation, we consider a very simple job shop first, in which each workstation consists of a single machine, cf. Fig. 19.2. The question now is to determine the network production rate (throughput) of the aggregated parts as a function of N , as measured by the load/unload station and denoted by TH0 (N ), as well as the lead time L(N ) which is the time between the moment a part is loaded at station 0 and its return to that station after completion

426

H. Zijm and M. Schutten

of its final operation, to be unloaded. Before introducing the Mean Value Analysis (MVA) algorithm in its simplest form, let us introduce some notation: Vm Qm (n)

visit ratio of station m, relative to the load/unload station 0. expected number of parts present at station m, when there are n parts in the network. lead time experienced at station m for an arbitrary part, when there are n Lm (n) parts in the network. THm (n) throughput at station m, when there are n parts in the network

Next, we formulate the MVA algorithm (which is exact for this specific situation) to calculate the throughput TH0 (N ) and the lead time L(N ). In fact, we will iteratively determine the TH0 (n) and lead time L(n) for any integer value n ≤ N , which will appear to be useful later. Mean value analysis algorithm for parts of one type in simple job shops 1. (Initialization) Set the visit ratio V0 to station 0 equal to 1. Determine the other visit ratio’s Vm , m  1, . . . , M , from the set of linear equations determined by the routing matrix, i.e., Vm 

M 

Vk Pkm

k0

Set n  0 and Qm (0)  0, m  0, . . . , M . 2. n : n + 1. 3. Compute Lm (n), m  0, . . . , M , from   1 Lm (n)  Qm (n − 1) + 1 μm 4. Compute L(n) and TH0 (n) from L(n) 

M 

Vm Lm (n)

m0

TH0 (n) 

n L(n)

and compute THm (n), m  1, . . . , M from THm (n)  Vm TH0 (n). 5. Compute Qm (n), m  0, . . . , M from Qm (n)  THm (n)Lm (n) 6. If n  N then STOP, else go to Step 2.

19 Advanced Production Planning and Scheduling Systems

427

Fig. 19.3 Throughput and lead time as a function of the number of parts in the system

Note that the MVA algorithm computes the system production rate (throughput) for any possible number of parts in the system. When displaying the calculated throughputs and lead times graphically, one may observe some remarkable phenomena (cf. Fig. 19.3). First, the system throughput asymptotically approaches a limit, which may be detected as the capacity of the bottleneck machine in the system, i.e., the machine m with the largest value of μVmm (hence the machine that spends the most time to a single part, taking into account the visit ratio). Clearly, with a low number of parts in the system, even a bottleneck machine may be temporarily idle, but at the same time, loading too many jobs in the system does not improve system throughput any further. The system lead time increases asymptotically linear with the number of parts in the system, since L(n)  THn0 (n) . Hence, one may conclude that the system should be sufficiently loaded to attain an acceptable throughput but not more since that would only increase the system lead time. This fundamental observation is the basis of the workload control concept—loading too many jobs in the system only increases the lead time and does not help to generate a significantly higher throughput. One may wonder why it is not possible in general to load the system such that exactly a bottleneck machine is always kept busy. For instance, a machine visited by every part exactly once, having an average processing time of 10 min, should be able to continuously work with an input flow of six jobs per hour. However, there are two reasons why this argument doesn’t hold. First, it should require a deterministic interarrival time of exactly 10 min at the machine as well. The latter cannot be guaranteed because other stations may be visited by the part before it visits this particular machine. It is even more difficult to ensure a deterministic interarrival time if the visit schemes differ among parts queuing up in front of a bottleneck machine. Second, an average processing time of 10 min does not mean

428

H. Zijm and M. Schutten

that every processing time is exactly 10 min. Apart from the aggregation discussed earlier, processing times are seldom deterministic; they usually vary due to operator, machine or material interruptions, set-ups required, quality problems, etc. On the other hand, if the number of parts loaded to the system is sufficiently high, there will be almost always parts buffered in front of the machine, preventing idleness of the bottleneck. The model and analysis presented above can be generalized in many ways. Workstations may contain multiple machines processing multiple part type families, each with their own (probabilistic) routing and processing time characteristics, including arbitrarily distributed processing times. In addition, extensions exist to include batching and set-up times (adding more variance to processing times, because typically the first part in a batch includes the set-up time in its process time), and quality problems causing possible rework loops (again a source of variability). Under these more general characteristics, the analysis is no longer exact but the procedures developed have been extensively tested and have proven to provide near-accurate estimates of important performance measures. The generalized MVA algorithms are mathematically complex and therefore beyond the scope of this chapter. Important however is that the general picture remains the same. The throughput increases with the number of parts loaded, but the function is concave. In other words, the law of diminishing returns applies. Beyond a threshold, additional parts loaded do not increase the throughput, instead lead to significant longer lead times.

19.5 Workload Control Under External Demand for Production to Stock (Advanced) So far, note that the CQN model is only used to determine the maximum capacity of a manufacturing department under a workload control regime (i.e., a regime that limits the number of parts released to the cell), consisting of various work stations that may be visited by various part types, each having their own routing through the cell. To that end, it was natural to assume that the department was always fully loaded. The parts manufacturing shop discussed in the introduction of this chapter however is subject to demand generated from a final assembly schedule. Because the variety of part types was limited in comparison to the large variety of end-items, parts are produced to stock (the CODP is located between parts manufacturing and assembly) to ensure that assembly is never idle due to a lack of parts. At the same time, a workload control regime limits the number of parts to be simultaneously processed to some number N . The question is how the preceding analysis can be used to accurately determine all performance measures of interest, including the fill rate of the finished parts inventories, the parts manufacturing lead times, and of course the resulting effective throughputs. In this section, we therefore consider a parts manufacturing department that produces items to be placed in stock, in anticipation of their use in a subsequent assembly

19 Advanced Production Planning and Scheduling Systems

429

phase. The parts manufacturing department is subject to a workload control regime. See Fig. 19.4, in which we restrict ourselves to one (generic) part type again to simplify explanation. In this system two control parameters play a crucial role, the predetermined stock level of finished parts S (called the base stock level) and the number of cards N . For each job (a part to be processed in the manufacturing shop), a card needs to be attached to guide it through the system. Hence, the number of cards N limits the number of jobs to be processed simultaneously. However, contrary to the capacity analysis in the preceding section, here not all cards are always in use; if demand from assembly shows a temporary decline, the production requirements diminish and hence a number of cards may be temporary idle. Now let m  the number of finished parts placed in stock, m ≤ S, k  the number of backordered demands for finished parts, a  the number of backordered production requests for parts, b  the number of free Kanbans available for parts production, b ≤ N . If a request for a part arrives from the assembly department, it is split into two subrequests. The first one is directed to the stock of finished parts m, at synchronization station Js . If a product is available in stock, it is used to satisfy demand, otherwise it joins a queue (k) of requests still waiting to be fulfilled (demand is backordered). The second sub-request reflects a production order for a similar part in order to make sure that the inventory position of finished parts is eventually replenished. However, the production order is released if and only if it is authorized by a card from the queue of free cards (b) at synchronization station Jc . If no free card is available, the production order joins a queue of orders that are waiting for a card and hence still

Fig. 19.4 A production-to-stock parts manufacturing shop controlled by a generalized Kanban control (GKC) system

430

H. Zijm and M. Schutten

have to be released. Finally, let n denote the vector with each element the number of parts at a workstation in the parts manufacturing department. Then, it is easily seen that: |n| + a + m − k  S |n| + b  N m·k 0 a·b0 Subtracting the first equation from the second equation and rearranging terms leads to (m − k)  (b − a) + (S − N ) Note that, if b − a is known, we know both b and a separately because both variables are nonnegative and cannot be non-zero simultaneously. If b − a  c then c ≥ 0 implies b  c and a  0, while c < 0 implies b  0 and a  −c. The same holds for m − k, in other words, both synchronization stations constitute a onedimensional queueing system, with states described by b − a and m − k respectively, which differ by a value S − N . Both queues act as a one-dimensional birth and death queue of which the state is increased by one due to a parts job completion, and is decreased by one due to an arrival of a finished parts request. Also observe that the parts manufacturing shop, together with synchronization station Jc , is in fact very similar to the closed shop in the preceding section, where station Jc takes the role of the load/unload station. The only, important, difference is that now a card that is dismissed after completion of a parts manufacturing job, is not immediately coupled to a new part to enter the shop, but may have to wait until all free cards ahead of it are occupied and then finally is attached to the next arrival for a parts production request. Suppose for the moment that the arrival of parts production requests is governed by a Poisson process with rate λ. Let c  b − a. Then the transition diagram between the states c (i.e., the birth and death process) is displayed as in Fig. 19.5, where TH (n) is the throughput of the parts manufacturing department treated as a CQN n cards circulating through the shop (as determined in the preceding section).

Fig. 19.5 Transition diagram of synchronization station Jc

19 Advanced Production Planning and Scheduling Systems

431

We now assume that the throughput rates TH (n) are exponentially and independently distributed (which is not true in general). This allows us to determine the probabilities p(c), as follows:   λN −c p(N ), c  N − 1, N − 2, . . . , 0, p(c)  N −c j1 TH (j)  −c   λN λ p(N ), c  −1, −2, . . . p(c)  N TH (N ) j1 TH (j) Finally, p(N ) follows as a normalization constant, i.e., from Nc−∞ p(c)  1. The number of parts n actually in process in the shop is now easily determined from n  N − max(c, 0) while also the probability distribution of m − k, and hence of m and k separately follows immediately from the probability distribution of c  b − a. From this, all relevant performance indices are now detected. As an example, we present the fill rate FR(S) of the stock of finished parts, i.e., the probability that an arriving request for a finished part is immediately fulfilled, when the base stock level is equal to S. We have FR(S) 

S 

p(m) 

m1

N 

p(c)

cN +1−S

while for the expected time to fill a part request ETFPR(S), the following equation holds. ETFPR(S) 

∞ 

(k + 1)p(k)/λ 

N −S 

(N − S − c + 1)p(c)/λ

c−∞

k0

The average lead time L∗ (N ) in the parts manufacturing shop satisfies L∗ (N )  L(N ) +

∞  a1

ap(a)  L(N ) +

−1 

(−c)p(c)

c−∞

where the first term on the right-hand side is the lead time as determined for a maximally loaded shop (as in Sect. 19.3) and the second term denotes the average waiting time before a card becomes available and hence production can be started. The above analysis can be extended to multiple part type families, each with their own routing. The analysis is dependent on whether one general workload norm for all part type families simultaneously holds (meaning one set of cards that limits the

432

H. Zijm and M. Schutten

overall workload) or a specific workload per part type family is specified (meaning a specific set of cards for each part type family). The analysis is beyond the scope of this chapter; we will not discuss it here any further. Further extensions to multi-stage systems in which each department is subject to a workload control system are also possible. Instead of an external demand modeled by a Poisson process with rate, we then encounter a state-dependent external demand rate (demand depends on the workload control rule, i.e., the number of cards, and the base-stock levels of a subsequent assembly section, whereas material availability in front of the parts manufacturing department again depends on base-stock levels and possibly a workload control rule at a preceding stage. We refer to the literature at the end of this chapter for further reading. Recall that in the entire analysis in this section we have assumed an FCFS priority discipline at each workstation, based upon which lead times have been determined. One may wonder whether at a detailed operational level smart scheduling systems might help to meet the due dates induced by these pre-determined lead times. That will be discussed in the final section.

19.6 Job Shop Scheduling (State-of -the-Art) In the preceding section, we determined the relation between the effective capacity and the expected lead time in the parts manufacturing shop. Based upon these expected lead times, we might set a due-date for each individual shop at the time of actual release to the shop floor. However, note that these due dates are then based on an average lead times and on a first-come-first-served priority discipline at individual workstations. In the current section we investigate whether at an operational level a more smart scheduling discipline can help to ensure that as many jobs as possible will indeed meet their scheduled due date. Such an advanced scheduling discipline exists. In this section, we outline the basic principles of what is known as the Shifting Bottleneck heuristic for a relatively simple job shop. Similar to the procedures in the previous section, also this heuristic can be extended rather easily to cope with generic versions of the job shop scheduling problem and is therefore suitable to be used in practice where all kinds of additional constraints may exist, both with respect to job as well as equipment constraints. Below, we first provide a formal description of the job shop scheduling problem that we need to solve. We then show how such problems can be represented by a disjunctive graph, after which we discus show to use the concept of selection to specify solutions for the job shop scheduling problem. Problem description In the job shop scheduling problem in its simplest version, a finite set of jobs J  {1, . . . , n} needs to be scheduled on M machines, numbered 1, 2, . . . , M . Each job j ∈ J consists of a number of operations that need to be processed in a specified sequence. For the sake of notational convenience, we assume that each job consists

19 Advanced Production Planning and Scheduling Systems

433

of M operations and that each job visits every machine exactly once, i.e., each job has one operation that needs to be processed on machine m (m  1, . . . , M ). Note that the sequences in which the jobs have to visit the machines may differ from job to job. Let Omj be the operation of job j that needs to be processed on machine m and let tmj denote the processing time. Moreover, let dj be the due date of job j, i.e., the time at which job j should be finished (as determined by its release time and the lead time as determined in the preceding section). Each machine can only process one operation at a time and once a machine starts processing an operation, it needs to finish processing this operation before it can start processing a next operation. Therefore, for each machine, we should find a sequence in which it processes its operations. A schedule σ specifies for each operation Omj its start time Smj (σ ) (i.e., the time at which machine m starts processing operation Omj ) and its completion time Cmj (σ )  Smj (σ ) + tmj (i.e., the time at which machine m completes operation Omj ). The maximum lateness of schedule σ is defined as   Lmax (σ )  max Lmj (σ )  max Cmj (σ ) − dj m,j m,j The objective is to find a schedule σ ∗ in which the maximum lateness Lmax (σ ∗ ) is as small as possible, which means that Lmax (σ ∗ )  min Lmax (σ ). σ

Disjunctive graph representation Each instance of the job shop scheduling problem can be represented by a graph G  (V, A), with V a set of nodes and A a set of arcs. V consists of a node vmj for each operation Omj and two dummy nodes, S1 (the source) and S2 (the sink). The weight of node vmj is equal to the processing time tmj of operation Omj . The weights of the nodes S1 and S2 are equal to 0. Suppose that Ohj and Omj are two consecutive operations of job j, which means that when job j has been processed on machine h, it needs to be processed next on machine Ohj and Omj , A contains an arc m. For each pair of consecutive

operations

vhj , vmj . Moreover, A contains an arc S1 , vmj if operation Omj is the first operation of job j. This means that there are n such arcs in total (one for each job). Finally, A contains n directed arcs from the last operation of each job to S2 . Together, these arcs form the conjunctive arcs of G. These conjunctive arcs represent the sequence in which the operations of each job need to be processed. Let C be the set consisting of these conjunctive arcs. Apart from the conjunctive arcs, A contains disjunctive arcs as well. These disjunctive arcs represent the machine conflicts. Because each machine can process at most one operation at a time, in a solution for the job shop scheduling problem we need to specify the sequences in which the machines process their operations. For every pair of operations Omj and Omk that be processed

on the same machine

need to m, A contains a pair of arcs, namely vmj , vmk and vmk , vmj , so one arc in both directions. Each arc has a weight of 0, except for the arcs to S2 . To model the job due dates, the arc from the node representing the last operation of job j to S2 gets a weight

434

H. Zijm and M. Schutten

Table 19.1 Data of an example instance

4

Jobj

dj

Omj tmj

1

17

O11 (4)

O21 (7)

O31 (6)

2

18

O32 (3)

O12 (5)

O22 (8)

3

14

O23 (2)

O33 (6)

O13 (7)

7

6 -17

3

5

8 -18 -14

2

6

7

Fig. 19.6 Disjunctive graph representation

equal to −dj . Figure 19.6 shows the disjunctive graph representation of a job shop scheduling instance shown in Table 19.1. Selection Now we have represented the job shop scheduling problem by a disjunctive graph. In this subsection, we subsequently show how a solution can be specified using this graph. Recall that the graph G contains a pair of arcs (one in each direction) between nodes vmj and vmk , representing two operations that need to be processed on the same machine m. A selection is a set of arcs that contains exactly one arc of each of these arc pairs. Let GS  (V, C ∪ S) be the graph that has the same node set V as the disjunctive graph G; moreover, the arc set of GS consists of the conjunctive arc set C of G plus the selection S. A selection S is called feasible if the corresponding graph GS is acyclic, i.e. if it does not contain a directed cycle. Recall that the objective in the job shop scheduling problem is to minimize the maximum lateness Lmax (σ ). This objective function is a regular objective function, which means that, given the sequences of operations on machines, it is always best to start each operation as early as possible. So, given a feasible selection S, we can determine its associated schedule σS by determining the earliest possible operation start and completion times, given the sequences of operations on machines implied by selection S. Figure 21.7 shows the graph GS for a feasible selection S and its associated schedule. Given a feasible selection S, the schedule σS associated with it can be calculated as follows. Let lmj (S) be the length of a longest path from S1 to vmj in the graph GS . The length lmj (S) is defined as the sum of the weights of the nodes and arcs that are part of the longest path from S1 to vmj in graph GS , excluding the weight of node vmj .

19 Advanced Production Planning and Scheduling Systems

435

Fig. 19.7 Graph GS and its associated schedule

0

4

7

0 3

4

0 2 0

6

5 4

11 8

-17 -18

6

11 7

3

9

-14

2

Fig. 19.8 Longest path calculations in graph GS

Given a feasible selection S, the earliest possible starting time of operation Omj is equal to lmj (S) (assuming that we begin the scheduling procedure at time 0, e.g. at the beginning of each week). This means that Smj (σS )  lmj (S). Moreover, Lmax (σS ) is equal to the length of a longest path from S1 to S2 . Figure 19.8 shows the longest path calculations for the same selection S as shown in Fig. 19.7 (resulting in the start times of all operations and in the maximum lateness Lmax (σS )). It is not difficult to see that there exists a selection S ∗ that results in an optimal schedule σ ∗ . Finding an optimal schedule (specifying the start and completion times of all operations) is therefore equal to finding the best selection (specifying the

436

H. Zijm and M. Schutten

operation sequences on machines, based on which the operation start and completion times can be calculated).

19.7 The Shifting Bottleneck Heuristic (State-of -the-Art) The Shifting Bottleneck (SB) is an iterative heuristic that decomposes the job shop scheduling problem into single-machine scheduling problems. In each iteration, one additional machine is scheduled until all machines are scheduled; the result is then a schedule for the job shop scheduling problem. In this section, we first discuss the Shifting Bottleneck heuristic and then zoom in on the required steps. The idea behind the SB heuristic is to focus first on machines that have most impact on the job shop schedule. The machine that is identified as having the most impact is called the first bottleneck machine and a schedule for this machine is determined. Next, in the second iteration, the consequences of the schedule of the first bottleneck machine is determined for each of the remaining machines. Then, of the remaining (non-bottleneck) machines, the machine with the most impact is determined. This machine is the next bottleneck machine and a schedule for this machine is fixed. At the end of the second iteration, the impact is determined that the schedule of second bottleneck machine has on the schedule of the first bottleneck machine. Possibly, the schedule of the first bottleneck machine is adapted to get a better overall result. This latter step is called the bottleneck reoptimization step and it ends the second iteration of the SB heuristic. In the third iteration, the third bottleneck machine is identified, its schedule is determined, and the bottleneck reoptimization step is performed. After the Mth iteration, the SB heuristic terminates. Before discussing the technical details of the steps, we first present a pseudocode for the steps at a high level: • • • •

//Initialization M :∅ //Execute M iterations to schedule all machines For m := 1 to M do – – – –

Find next bottleneck machine (bm ) Fix schedule of new bottleneck machine bm Reoptimize the (m − 1) existing bottleneck machines M : M ∪ {bm }

We now proceed to discuss the technical details of the steps in the Shifting Bottleneck heuristic. Find next bottleneck machine and its schedule At the start of iteration m of the SB heuristic, the schedules of (m − 1) bottleneck machines have been fixed. For these machines, we therefore have sequences in which these machines process their operations. In other words, we have a partial G S . If we ignore the capacity constraints of selection S . Consider now the

graph non-bottleneck machines, rmj S := lmj S gives the earliest possible starting time

19 Advanced Production Planning and Scheduling Systems

437

(i.e., the release date) of each operation Omj , while the length of

a longest path in GS S be the length of a from S1 to S2 gives the maximum lateness Lmax (σS ). Let lmj longest path in GS from vmj to S2 . If we do not want to increase the current maximum

lateness Lmax (σS ), then operation Omj should start no later than Lmax (σS )−lmj S .



Therefore, dmj S := Lmax (σS ) − lmj S + tmj is a due date for operation Omj . in GS , we have an earliest possible starting time

Based on longest path calculations

rmj S and a due date dmj S for each operation Omj . For each of the non-bottleneck machines, we now determine the optimal (single-machine) schedule, taking into account the release and due dates by means by an algorithm derived by Carlier (1982); the objective in these single-machine scheduling problems is also to minimize the maximum lateness. The non-bottleneck machine that has the highest maximum lateness is the next bottleneck machine. The schedule that is fixed for this machine is the schedule that results in the optimal maximum lateness. Reoptimize the existing bottleneck machines Suppose again that we are at the mth iteration of the SB heuristic and, moreover, that we just identified the mth bottleneck machine. To finalize the mth iteration of the heuristic, we now perform the bottleneck reoptimization step. Suppose that bk is the bottleneck machine found in the kth iteration (k  1, . . . , m). During the bottleneck reoptimization step, the m − 1 existing bottleneck machines are rescheduled one by one, taking into account the schedules on the other bottleneck machines, including the one identified in this (mth) iteration. Once an existing bottleneck machine is rescheduled, the newly found schedule replaces the existing schedule for this bottleneck machine. To find a new schedule for bottleneck machine bk , consider the partial selection Sk that consists of the newly found schedules for machines b1 , b2 , . . . , bk−1 and the schedules for machines bk+1 , bk+2 , . . . , bm . For operations Obk ,j on bottleneck machine bk , the release and due dates are calculated in the graph GS , again based on k longest path calculations. The new schedule for machine bk is the schedule on this machine that minimizes the maximum lateness given the release and due dates. In pseudocode, the bottleneck reoptimization procedure reads as follows: • • • •

//at the start, we have the existing bottleneck //machines b1 , b2 , . . . , bm−1 and //a new bottleneck machine bm For k := 1 to m − 1 do – Construct GS k – Calculate release and due dates for operations on machine bk – Find optimal schedule for machine bk .

Practical extensions One of the strengths of the SB heuristic is the possibility to modify it such that it can be applied to far more general versions of the job shop scheduling encountered in practice. The modifications typically consist of (a combination of) changes in the

438

H. Zijm and M. Schutten

weights of the nodes and arcs in the disjunctive graph and using adapted algorithms to solve the single-stage scheduling problems (which are single machine scheduling problems in the version described above). In this way, the SB heuristic is for example able to deal with setup times on machines, or workstations that consist of parallel identical machines instead of a single machine.

19.8 Further Reading In this chapter, we have presented algorithms for executing important building blocks of the framework presented in Chap. 5 and present the basic structure of a (metal working) company as often exists in practice. Such a company is detailed in the case study presented in Sect. 21.2, which is derived from Slomp (1993). The linear programming formulations discussed in Sect. 21.3 uses elements from various publications, including Silver et al. (2017), Bitran et al. (1982) and Hopp and Spearman (2008). Interesting also are attempts to integrate MRP and HPP, see e.g. Hax and Meal (1975) and Hax and Candea (1984). The relation between workload, throughput (effective capacity) and internal lead times in a manufacturing shop are discussed by Hopp and Spearman (2008), but the complete technical basis is due to Buzacott and Shanthikumar (1993), using a queueing network approach. This work (and the papers it is based on) led to a rich offspring, see e.g. Dallery (1990), Di Mascolo et al. (1996), and Buitenhek et al. (2000), while an attempt to place the work in a broader manufacturing planning and control context was presented by Zijm (2000). Zijm and Buitenhek (1996) investigate the integration of a due-date setting approach based on (open) queuing networks and the subsequent application of a Shifting Bottleneck heuristic. Job Shop scheduling is a topic that has received much attention in combinatorial optimization of which the book by Pinedo and Chao (1996) presents a nice review, see also Brucker et al. (1994). The Shifting Bottleneck procedure was introduced by Adams et al. (1988), using Carlier’s one-machine scheduling problem as a building block (Carlier 1982), and generated quite some offspring, see e.g. Ivens and Lambrecht (1996). Extensions that make the procedure applicable to more realistic machining systems are due to Schutten (1998), Schutten and Leussink (1996) and Schutten et al. (1996). The integration of workload control, lead time off-setting and shop floor scheduling remains a challenging task. The wealth of generic heuristics for combinatorial optimization problems (simulated annealing, taboo search, genetic algorithms) have contributed significantly to deriving close-to-optimal solutions for these problems. At the same time, they do not provide the insights that can be expected from a further integration of queueing and smart scheduling approaches. The latter defines a rich field for future research.

19 Advanced Production Planning and Scheduling Systems

439

References Adams J, Balas E, Zawack D (1988) The shifting bottleneck procedure for job shop scheduling. Manag Sci 34:391–401 Bitran GR, Haas EA, Hax AC (1982) Hierarchical production planning: a two stage system. Oper Res 30:232–251 Brucker P, Jurisch B, Sievers B (1994) A branch and bound algorithm for the job-shop scheduling problem. Discrete Appl Math 49:107–127 Buitenhek R, van Houtum GJ, Zijm WHM (2000) AMVA based solution procedures for open queueing networks with a population constraint. Ann Oper Res 93:15–40 Buzacott JA, Shanthikumar JG (1993) Stochastic models of manufacturing systems. Prentice-Hall, Englewood Cliffs, NJ Carlier J (1982) The one-machine sequencing problem. Eur J Oper Res 11:42–47 Chen H, Yao DD (2001) Fundamentals of queueing networks—performance, asymptotics and optimization. Springer, New York Dallery Y (1990) Approximate analysis of general open queueing networks with restricted capacity. Perform Eval 11:209–222 Di Mascolo M, Frein Y, Dallery Y (1996) An analytical method for performance evaluation of Kanban-controlled production systems. Oper Res 44(1):50–64 Hax AC, Candea D (1984) Production and inventory management. Prentice-Hall, Englewood Cliffs, NJ Hax AC, Meal HC (1975) Hierarchical integration of production planning and scheduling. In: Geisler MA (ed) Logistics. Studies in the management sciences, vol 1. Elsevier, North-Holland Hopp WJ, Spearman ML (2008) Factory physics: foundations of manufacturing management. Richard D. Irwin, Homewood, IL Ivens P, Lambrecht M (1996) Extending the shifting bottleneck procedure to real-life applications. Eur J Oper Res 90:252–268 Pinedo M, Chao X (1996) Operations scheduling with applications to manufacturing and services. Irwin/McGraw-Hill, Boston Reiser M, Lavenberg SS (1980) Mean-value analysis of closed multichain queueing networks. J Assoc Comput Mach 27:313—322 Silver EA, Pyke DF, Thomas DJ (2017) Inventory and production management in supply chains. CRC Press (Taylor and Francis), Boca Raton Schutten JMJ (1998) Practical job shop scheduling. Ann Oper Res 83:161–177 Schutten JMJ, Leussink RAM (1996) Parallel machine scheduling with release dates, due dates and family setup times. Int J Prod Econ 46–47:119–125 Schutten JMJ, van de Velde SL, Zijm WHM (1996) Single-machine scheduling with release dates, due dates and family setup times. Manag Sci 42(8):1165–1174 Slomp J (1993) Production control for flexible manufacturing system—an application-oriented approach. PhD thesis, University of Twente Zijm WHM, Buitenhek R (1996) Capacity planning and lead time management. Int J Prod Econ 46–47:165–179 Zijm WHM (2000) Towards intelligent manufacturing planning and control systems. OR Spectr 22:313–345

Chapter 20

Stochastic Inventory Models Henk Zijm

Abstract We discuss inventory systems in an independent demand setting, where demand over time is modeled as a stationary stochastic process. We begin with some basic notions and definitions on inventory management, followed by a discussion of well-known (and applied) control systems. Under periodic review and a linear cost structure, it is known that the optimal control policy has a critical level structure, hence we analyze such critical level policies in detail. After that, in an advanced section, we turn to multi-echelon or multi-stage systems. We present a complete analysis of the decomposition result proven initially by Clark and Scarf, and its analogue in distribution systems, i.e., systems with an arborescent instead of a linear structure (state-of-the-art). Computational aspects are briefly discussed after which we close with some guidelines for further reading.

20.1 Stochastic Inventory Models: Definitions and Terminology (Basic) Inventories are everywhere. As far back as ancient nomadic and agricultural civilizations, mankind has been storing products for future use. Indeed, storage of agricultural products is required to bridge the time between the (often short) harvest season and future consumption. In modern industrial organizations, inventories are indispensable to match supply and demand, for various reasons. The most common reason to store products can be attributed to the uncertainty in future demand. While organizations would like to minimize the risk of being unable to fulfill customer demand and loose both goodwill and revenue, they also like to keep their inventory costs low. Thus, it is necessary to have some level of inventory in many organizations. Upfront, inventories may also serve as a protection against quality defects or supply uncertainty, e.g., due to malfunctioning production equipment. Capacity limitations H. Zijm (B) Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, Overijssel, The Netherlands e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_20

441

442

H. Zijm

are another reason for storing products in anticipation of future needs. Note that adjusting capacity is difficult in many industries, especially those in which the cost to switch from one product to another are high. If demand is highly volatile and capacity is set to match average demand, then advance production to meet future peak demands is necessary, again resulting in inventories. This is especially true for seasonal products. The need to minimize production costs, e.g., of large batches for products that have high production set-up times or costs, leads to a build-up of inventories. These inventories are gradually consumed and upon their depletion, a new batch is produced. In all cases, inventories basically represent an alternative form of (allocated) capacity. However, inventories also represent capital invested which therefore cannot be used other than to fulfill future demand. The Economic Production Quantity discussed in Chap. 12 appeared to be a first attempt to balance production set-up and inventory costs, in a deterministic demand setting. Such a deterministic scenario is common in a dependent demand situation in which the needs for parts and components are derived from a master production schedule at end-item level, as in an MRP driven production environment. When demand is independent, i.e., generated by external sources, it is prudent to model it as a stochastic process. That is the focus of the current chapter. Inventory modeling, planning and control is a topic that has received significant attention in the Operations Research and Operations Management literature. The pioneering publication of Arrow et al. (1958) marked the start of an extremely fruitful period in the study of stochastic inventory models, with an emphasis on structural properties of optimal control policies. However, many models turned out to be computationally intractable, which has led to a range of heuristic approaches. The number of books on the topic is overwhelming and we will not make an attempt to review them. Instead, we only mention some important results that still provide a basis for further study. In this chapter, we limit ourselves mainly to periodic review systems with full backlogging that either consist of a single stage or involve multiple stages, mostly under centralized control. We begin by introducing some basic definitions and terminology. Consider a planner in a warehouse who is responsible for having sufficient items in stock to fulfill anticipated future demand. Typically, the planner will have an idea on what demand can be expected in some time period ahead. He or she can order items from an external supplier if the available stock is about to be depleted, thus building inventory to meet future demand. Before we describe the multiple inventory policies a planner may use, we begin with some important definitions, cf. Silver et al. (2017). A stock keeping unit (SKU) refers to one particular item or product, characterized by attributes such as size, color, and function. The on-hand inventory (OH) of an SKU refers to the quantity of that SKU physically present, i.e., available on the shelf and to be used to satisfy any customer demand. The net stock (NS) of an SKU is defined as the on-hand inventory minus the number of backorders. Note that under reasonable circumstances, there is either on-hand inventory or there are backorders, denoted by BO. Thus, NS  OH −BO.

20 Stochastic Inventory Models

443

The inventory position (IP) of an SKU equals the on-hand inventory, plus the number of items ordered that have not yet physically arrived, minus the number of backordered items. The (supply) lead time of an order for a particular SKU is the time that elapses between the time when the order was placed and the time when the order arrives. In inventory theory, we discern two basic situations on how to handle a customer’s order when the desired item is temporarily not available in stock. In the case of backordering, it is assumed that the customer is willing to wait until the item becomes available again. Upon arrival of an order (usually a batch of items), backorders are fulfilled on a first-come, first-served (FCFS) basis. The alternative is lost sales, in which case it is assumed that the customer seeks another vendor to satisfy his or her demand. In general, however, a sales organization may face a mix in which some customers are willing to wait and others go elsewhere, while also some products lend themselves more easily for backordering than others (e.g. when they are unique in some sense). Note that the occurrence of a lost sales is not always easily known, because customers may not inform a vendor what they actually do if their demand is unmet by that vendor. They may purchase the product from an alternate vendor, opt for demand substitution by purchasing another product that is reasonable close in functionality to the one they were seeking from the original vendor or a competitor, or just may decide not to buy the product at all. In what follows, we will restrict our discussion to inventory models with full backordering. Generally, an inventory planner will attempt to fulfill demand even if it is higher than expected. That is the function of safety stock, which typically is meant to deal with random demand fluctuations. In the case of smooth, fully predictable demand there is no need to reserve safety stock but when demand fluctuates over time, it may play an essential role in preventing stock-outs. The latter result in either backorders or lost sales and hence in both cases, a reduced customer service level, or additional costs, or both. We will come back on how to model possible penalty costs in the case of a stock-out, and to various definitions of a customer service level next. We now discuss when to monitor the inventory status and possibly to place a new order at an upstream supplier or manufacturing department. Before doing so, it is important to determine whether we face a stationary or a non-stationary demand. A stationary demand process is not constant but is modeled as a stochastic process with a fixed mean and variance over time, whereas the mean and variance of a nonstationary demand processes change over time (e.g., in the case of seasonal demand patterns). In this chapter, we assume that demand is stationary with known mean and variance. Modern warehouse management systems allow for a continuous review and update of the inventory position. Thus, in principle, orders can be placed continuously. However, many organizations choose to place orders periodically, i.e., after one or more time intervals of fixed length. In the retail sector, periodic review and ordering is common practice, simply because a large number of different SKU’s can then be ordered simultaneously. The choice of the length of the time intervals must be determined, but often follows naturally from the calendar or from distribution schedules

444

H. Zijm

of logistics service providers, leading to the possibility of placing a replenishment order periodically, e.g., once a day or once per week. When reviewing stock, it seems natural to order a replenishment quantity if the inventory position is so low that a further postponement of a replenishment order may lead to a stock-out in the near future. Following this argument, often a minimum inventory level s is chosen such that s may cover the expected demand plus a certain fraction of the demand variation during the time it takes for the new order to arrive and become available to fulfill future demand. The determination of this so-called re-order point s will be discussed in more detail below for a periodic review system. In addition to the determination of the reorder point, which determines when to order, a companion question is: how much to order? Often, ordering costs consist of two components, a fixed cost, which is independent of the order size, and variable purchasing costs that are a linear function of the order size. Because of the fixed costs, it makes sense to order at least a minimum quantity, while on the other hand large orders may lead to high inventory holding costs (cf. the discussion on the Economic Order Quantity in Chap. 12). Two systems are often distinguished: one in which a replenishment order has a fixed size of Q, say, and one in which the inventory position after ordering is brought to a fixed order-up-to level S, say. Combined with the continuous versus periodic review options, this leads to four system control categories: Continuous review (s, Q) systems: As soon as the inventory position drops below a specified re-order point s, a replenishment order of size Q is placed. The order will arrive after L time units. If every customer requires exactly one unit, the reorder will be placed when the inventory position equals s. However, when a customer demands more than one unit, the IP may drop instantaneously below s. Continuous review (s, S) systems: As soon as the inventory position drops below a specified re-order point s,a replenishment order is placed such that the inventory position is returned to an order-up-to level S. The order will arrive after L time units. Periodic review (R, s, Q) systems: At time points t  0, R, 2R, . . . , the inventory status is reviewed. As soon as the inventory position is equal to or smaller than s at a review point t, an order of size Q is placed, which will arrive at time t + L. Periodic review (R, s, S) systems: At time points t  0, R, 2R, . . . , the inventory status is reviewed. As soon as the inventory position is equal to or smaller than s at review point t, an order is placed such that the inventory position is returned to an order-up-to level S. The order will arrive at time t + L. It is important to realize that the variable to be observed is the inventory position IP, not the net stock. In particular, if more than one replenishment order is underway, it is important also to keep track of items that may arrive soon, instead of only the net stock. Also, note that if the inventory position just before ordering is i ≤ s, and an order of size V is placed, upon arrival of that order, the resulting inventory position will in general be less than i + V , due to the demand that was satisfied from the currently available stock after the placement of the order. In particular, in the case of an (R, s, S) system, the inventory position just after arrival of a placed order will in general be less than S (even if only one replenishment order is issued at a time). See Fig. 20.1.

20 Stochastic Inventory Models

445

Fig. 20.1 A single-item periodic review (R, s, S) inventory system

In a periodic system we generally assume that the parameter R is determined in advance, e.g., as a day or week, so we need to determine the re-order point s and either the order quantity Q or the order-up-to level S, depending on what policy is applied. Hence, in any system two parameters are to be determined. In the case of a periodic system, when we take s  S, this means that at each review moment, the observed inventory position i is increased with a replenishment order of size S − i. Such systems are generally denoted as (R, S) systems, in which only the parameter S has to be determined. These systems typically occur in large retail stores that periodically place replenishment orders for a large number of different SKU’s, that are delivered simultaneously. It may happen in non-retail (e.g., manufacturing) industries as well, however, note that in the case of larger lead times L, the likelihood of two or more replenishment orders for a similar SKU being underway at the same time increases. Finally, we discuss cost parameters. As previously mentioned, there is a fixed cost K associated with the placement of each replenishment order (independent of its size) while in addition a variable cost c is incurred for each item purchased. For each item of a particular SKU that is stocked, an inventory holding cost h is incurred which generally equals a fraction r of the variable purchasing cost of that item, hence h  rc. Although depending on the type of industry, a fraction r  0.2 (20%) is not uncommon in inventory management; generally such a percentage does not only cover the interest rate, but also the costs of storage (often based on the depreciation of an overall investment in warehouse building and equipment), the operational costs of all materials handling and support activities, insurance costs, opportunity costs, and obsolescence risks. In case a stock-out occurs, typically shortly before a replenishment order arrives, one may in addition incur a penalty cost p per item short per time unit. Such costs may indeed represent physical costs (e.g., costs of a crash action or, for example in the spare parts business, the costs of downtime of a client’s system while the inventory management is contractually obliged to deliver a desired part immediately upon request). Often, the penalty costs also serve to enforce a desired customer service level, e.g., defined as the fill rate, i.e., the number of items

446

H. Zijm

delivered from the shelf (hence without backorders). It is intuitive, and it can be proven that higher penalty costs induce a higher customer service level and vice versa, cf. Van Houtum and Zijm (2000). In fact, there are several ways to penalize a shortage, and also several ways to define customer service levels. See Silver et al. (2017) for an overview, and Van Houtum and Zijm (2000) for the relations between various penalty functions and service levels. Now, to find the right parameters in any of the four control categories above, one may proceed as follows. In case penalty costs are specified, a planner generally attempts to minimize the sum of the ordering, inventory holding, and penalty costs, averaged over a sufficiently long horizon. When it is difficult to assess a penalty cost, a target customer service level (CSL) is often specified. In this case, the planner may attempt to minimize the sum of the ordering and inventory holding costs, averaged over a sufficiently long horizon subject to the constraint that on average the requested service level is met. There are several direct approaches for the latter problem, but alternatively one may solve a pure cost problem with some artificial penalty costs, measure the optimal policy’s customer service level and next adjust the penalty cost after which the procedure is repeated. Because the CSL is generally increasing as a function of p and vice versa, a simple bisection procedure is sufficient to determine the right penalty costs. Van Houtum and Zijm (2000) have proven that an optimal policy for the pure cost formulation, which results in a specific CSL, is also optimal for the problem in which order and inventory-holding costs are minimized, subject to a service level constraint based on the same CSL. In practice, it is not always easy to determine the right parameters numerically. Fortunately, a number of structural properties can be exploited. The most important one is that for periodic review problems with a cost structure as defined (with a penalty cost p per unit short per period), the optimal policy turns out to belong to the category of (R, s, S) policies, see Scarf (1960). Because periodic review systems are generally used in practice, we restrict ourselves in this chapter to an analysis of (R, s, S) systems and a reduced version, i.e. (R, S) systems.

20.2 Case Study: Inventory Management at IKEA IKEA is one of the world’s largest home furnishing companies. Its name is an abbreviation of “Ingvar Kamprad Elmtaryd Agunnaryd”, referring to the founder Ingvar Kamprad who grew up on the farm Elmtaryd in the town of Agunnaryd in Southern Sweden. Kamprad started the company in 1943, at the age of 17. Today, IKEA owns and operates more than 400 retail stores in 49 countries. Good quality products at low prices is the motto of the company. The vision of IKEA is to provide well designed, functional home furnishings at prices so low that as many people as possible will be able to afford them (www.ikea.com). To realize that vision, the various supply chain functions (procurement, inventory management, sales) are carefully tuned, in that way

20 Stochastic Inventory Models

447

contributing to the company’s strong competitive position. Most IKEA products are procured from external suppliers, while the SWEDWOOD group, an IKEA subsidiary with its largest factory in Southern Poland, is responsible for the manufacture of the entire set of wooden products. One of the principles of IKEA is to commit to a catalog of products that are stocked for a year at a guaranteed price. Another distinctive feature is that most furniture is not sold pre-assembled, but designed to be self-assembled. IKEA’s well-known wooden products are characterized by a high degree of modularity. The standard elements (modules) can be combined in an almost unlimited number of variants, allowing for a high level of customization while remaining cost effective. Modularity and standardization are indeed key elements of IKEA’s strategy. IKEA’s retail stores are megastores, with a design that defines pathways that guide the customer in a natural way along all the functions that are part of home living. These paths end in a self-service warehouse where customers themselves pick up the requested modules (unless they are too heavy for manual handling). Storage columns in the warehouses are separated in a retrieval area (downstairs) and a bulk area (upstairs) that serve to replenish the retrieval area overnight. Customers can order and purchase products from IKEA in three different ways, at retail stores, by phone or via the internet. In the latter two cases, the products are delivered directly to the customer’s home, from a customer distribution center and often via a local hub. Some of the articles are replenished to stores from distribution centers while others are delivered directly to stores from suppliers. Direct deliveries minimize the handling costs as well as transportation cost, but on the other hand generally drive up the stock levels. Articles delivered through distribution centers are mainly divided into fast and slow movers. The fast mover distribution centers are either used for storage or as transfer centers where articles are repacked and shipped to retail stores. Lead times from suppliers to distribution centers and stores are normally a few weeks while lead times from distribution centers to retail stores are only a few days. Slow movers are articles subject to low volume are stored in only a small number of distribution centers, each one supplying an entire market with slow moving articles. The idea is to create economies of scale and therefore reduce costs by avoiding many small inventories across distribution centers. The in-store logistics managers use an inventory replenishment management process developed by IKEA called ‘minimum/maximum settings’ where ‘minimum’ refers to the minimum amount of products available before reordering, and ‘maximum’ defines an upper limit on the number of a particular product to order at one time. Still, IKEA feels that the inventory levels kept at retail stores and distribution centers throughout the company are generally high. In addition, inventories at the distribution centers and the retail stores are controlled independently and safety stocks should be sufficiently large to cover

448

H. Zijm

uncertainties from the next level of demand. In other words, the various echelons are not synchronized and the system as a whole may be considered to operate suboptimally. Service level requirements are the same for both retail stores and distribution centers, while also IKEA management realizes that the service level experienced by customers is the only one that matters. The two questions that IKEA faces are: what is an optimal inventory control policy, and does a coordination of inventory policies across retail stores and distribution centers help reduce overall inventory?

20.3 Periodic Review Stochastic Inventory Systems with Backlogging (Advanced) In this section, we analyze periodic review stochastic inventory systems in which we seek the policy that minimizes the average inventory holding and ordering costs, subject to a service level constraint. A fixed order cost K is incurred whenever a replenishment order is placed, while in addition there is a variable cost c for each item purchased. In addition, we incur an inventory holding cost h  rc per item per period. For convenience, we assume that stock is monitored at the end of each period, while ordering takes place at the beginning of a period. Orders also arrive (and its SKU’s are ready to be used) at the beginning of a period. Demand arises throughout any period and is modeled by a random variable that denotes the total demand in such a period. Because we assume a stationary demand process, we can denote the random periodic demand variable as uR , with mean μR and standard deviation σR . We may choose uR to be either a discrete or a continuous random variable. Although discrete variables may be more natural, they also give rise to cumbersome notations and computational problems, therefore we have chosen here to model demand as a continuous random variable, with pdf FR and density function fR . A fundamental equation in many stochastic inventory models is the so-called newsvendor equation (see e.g. Silver et al. 2017) which will also form the start of our discussion. The name refers to the problem of a newsvendor who wonders how many newspapers to purchase at the beginning of a day. He or she earns an additional revenue on each newspaper sold but has to deal with unsold newspapers at the end of the day, which will incur disposition or salvage costs. Some reflection shows that this problem is similar to the determination of the inventory holding and penalty costs at the end of a period. At the beginning of the period, the total net inventory of an SKU is brought up to some quantity, say y, and demand is satisfied as much as possible during the period. Recall that in an inventory system with backordering, the net inventory may be negative (which is not realistic for the newsvendor problem). In particular, let

20 Stochastic Inventory Models



y

H (y)  h

449

 (y − u)fR (u)du + p

0



(u − y)fR (u)du,

y ≥ 0,

y

 H (y)  p



(u − y)fR (u)du,

y < 0,

0

where, as before, fR denotes the density function of demand in a period of length R. Now, assume that at the beginning of the period the net stock equals x and that we wish to bring the stock up to a level y, which takes effect immediately (hence, with zero lead time). If a fixed cost K is incurred when placing an order (next to the variable costs for each product), the total expected costs over the period are equal to K + c(y − x) + H (y), if y > x, and H (x), if y  x (that is, if we order nothing). Note that H (y) is a strictly convex function of y and hence the same holds for cy + H (y). The following derivation closely follows Ross (1970). Let cy+H (y) be minimized in y  S and defined s as the smallest value for which cs + H (s)  K + cS + H (S) Now, we distinguish the following three cases: x > S: then cy + H (y) > cx + H (x) for all y > x, hence K + c(y − x) + H (y) > H (x) for all y > x. Therefore, it is optimal not to place an order. s ≤ x ≤ S: then K +cy +H (y) ≥ cx +H (x) for all y > x, hence K +c(y − x)+H (y) ≥ H (x). Once again, it is optimal not to place an order. x < s: then miny≥x {K + cy + H (y)}  K + cS + H (S) < cx + H (x), hence min{K + c(y − x) + H (y)}  K + c(S − x) + H (S) < H (x) y≥x

Therefore, in this case it is optimal to bring the inventory up to level S. Now, for an n-period inventory problem, the total costs Cn (x) in a discounted cost framework satisfies the following dynamic programming recursion:    ∞ Cn−1 (y − u)fR (u)du (20.1) Cn (x)  min K∂(y, x) + c(y − x) + H (y) + α y≥x

0

where ∂(y, x)  1 if y > x and ∂(y, x)  0 if y  x, and α < 1 is a periodic discount factor. Scarf (1960) initially showed, using a Dynamic Programming formulation, that for a stationary demand process, the optimal control policy under a discounted cost framework in each period n is of the type (sn , Sn ). Using these results, Iglehart (1963) demonstrated that the sequence (sn , Sn ) converges to two fixed values (s, S) for n → ∞. Thus, in an infinite horizon problem, the same (s, S)-policy appears to be optimal in each period among all possible control policies. The same result holds when switching to an average cost criterion by considering the limiting behavior of Cn (x)/n for n → ∞ and letting α → 1.

450

H. Zijm

The result also remains valid when assuming positive lead times L, where for convenience we assume that L is a multiple of R (note that this is not really a restriction because R can be set to a small value, e.g., equal to one day). Basically, one needs to keep track of not only the current net stock position x but also the goods that will arrive in the next L − 1 future periods x1 , x2 , . . . , xL−1 (based on decisions taken L−1 in the past), thereby leading to an L-dimensional state space. Note that x + j1 xj equals the inventory position. Any decision to order an amount of goods z at the beginning of period t will take effect only when the order actually arrives at the beginning of period t + L. Using a reduction argument, Scarf (1960) showed that the dynamic programming recursion for the corresponding multi-variable costs Cn (x, x1 , x2 , . . . , xL−1 ) can bereduced to arecursion similar to (20.1) for a related  one-dimensional function Cˆ n x + L−1 j1 xj , in which H (y) is replaced by a more complicated L-fold integral function. Using this reduction, it is then easily shown that the optimal policy in the stationary infinite horizon problem again is of the (s, S)type, where the variable under consideration is the inventory position xˆ  x+ L−1 j1 xj , i.e., if xˆ ≤ s we order S − xˆ , while if xˆ > s, we order nothing. The structural result on the optimality of an (s, S)-strategy for all possible control policies is extremely important but unfortunately, the calculation of the parameters s and S is a bit complicated, although various authors have developed algorithms that allow a fast computation (see e.g., Zheng and Federgruen 1991, who study the equivalent discrete state space problem). Below, we follow another approach which views the problem from a joint cost and service based perspective, instead of considering only a pure cost approach. Although we make slightly different choices when selecting the parameters s and S, our analysis resembles the one developed by Tijms and Groenevelt (1984). See also Silver et al. (2017). Instead of minimizing the sum of the ordering, inventory holding, and penalty costs, we now consider an infinite horizon stationary demand problem in which we wish to minimize the average expected ordering and inventory holding costs, subject to a service level constraint. (Note that the average purchasing costs are constant and hence can be ignored in any optimization procedure). More precisely, we wish to consider policies which guarantee a long term average fill rate β, where the fill rate is defined as the fraction of demand that can be satisfied immediately from the available stock. In other words, the fraction of demand to be satisfied from physical stock should be larger than or equal to β in the long run, while, as before, all unsatisfied demand is backlogged. Again, by equivalence it can be shown that the optimal policy for this problem is of the (s, S)-type. Now, we turn to the computation of the control parameters. It may be helpful to review the role of the fixed ordering cost K in relation to the inventory holding cost h. Typically, one may expect S − s to increase as a function of the ratio K/h, as also argued in the discussion on the Economic Order Quantity in Chap. 12. Note however that, because we observe the inventory position only at discrete points in time, the order size will not be fixed but equal S −s+z where z is the so-called undershoot, i.e., the difference between the reorder point s and the actually observed inventory position at the time of re-order. The choice of the re-order point

20 Stochastic Inventory Models

451

s should reflect the target customer service level or fill rate β and should take into consideration the expected undershoot. By analogy it is easily shown that the demand process in subsequent periods is a renewal process (cf. Ross 1970), i.e., a sequence of independent identically distributed random variables un with common distribution function FR . For such a renewal process it is well known that in the long run, if one picks an arbitrary point in time, the expected time t until the next renewal (often called the residual life time) has a density function equal to μ1R (1 − FR (t)). Some reflection shows that in our inventory model the undershoot z exactly behaves like the residual lifetime in the analogous renewal process. Therefore, if S − s is sufficiently large (hence if K/h is large), the density function fz of the undershoot z approximately satisfies fz (z) 

1 (1 − FR (z)) μR

From the above, it is easy to show using partial integration, that the mean μz of z equals ∞ μz  0

1 zfz (z)dz  μR

∞ 0

1 z(1 − FR (z))dz  2μR

∞ z 2 fR (z)dz 

σR2 + μ2R 2μR

0

where σR2 denotes the variance of the periodic demand. At the end of this section we discuss what to do if K/h is not ‘sufficiently’ large, i.e., if K is small or even zero. σ 2 +μ2 For now, if we decide to order, the average order size will be equal to S −s+ R2μR R . Therefore, it seems reasonable, following the EOQ derivation, to make  σR2 + μ2R 2KD S −s+  2μR h where D is the expected annual demand, i.e. D  (Ndemand /R)μR , Ndemand is the number of “demand days per year” (not necessarily equal to 365) and R is expressed

in days. However, this does not mean that an order always equals 2KD , but merely h that, once we know how to calculate s, the order-up-to level S follows immediately. As mentioned previously, once we order, say at time t, the net stock position is influenced only L periods later, i.e., at the beginning of period t + L. Hence, the remaining stock s − z at time t should be sufficient to cover demand in the next L periods. In other words, at the end of period t + L − 1 (just before the order arrives), the net stock is equal to s − z − uL , where uL denotes the demand during the lead time L. Note that we either observe some positive net stock (equal to the on-hand inventory OH ) or a negative net stock (i.e., a backorder position BO). Let x  z+ uL , with density function fx , then we may compute the means μOH and μBO at the end of period t + L − 1, again by using partial integration, as follows:

452

H. Zijm

s

s

μOH 

(s − x)fx (x)dx  0

1  μR 

1 μR



s

(s − x)⎣

0

⎡ (s − x)⎣

0

x

x 0

⎤ fz (x − u)fL (u)du⎦dx ⎤

(1 − FR (x − u))fL (u)du⎦dx

0

s (s − x)(FL (x) − FR+L (x))dx 0

⎡ s ⎤  s 1 ⎣  (s − x)2 fL (x)dx− (s − x)2 fR+L (x)dx⎦ 2μR 0

0

and ∞ μBO 

∞ (x − s)fx (x)dx 

s







1 μR 1 μR 1 μR

s

∞

0

⎤ ⎡ x  (x − s)⎣ (1 − FR (x − u))fL (u)du⎦dx

s

0

∞ (x − s)(FL (x) − FR+L (x))dx s

∞ (x − s)[(1 − FR+L (x)) − (1 − FL (x))]dx s





⎤ ⎡ x  (x − s)⎣ fz (x − u)fL (u)du⎦dx

1 ⎣ 2μR

∞

∞ (x − s)2 fR+L (x)dx−

s

⎤ (x − s)2 fL (x)dx⎦

s

The safety stock SS is generally defined as the net stock just before an order arrives, i.e., SS  s − μz − μL  s −

σR2 + μ2R − μL 2μR

The reader may easily verify that SS  μOH − μBO . The re-order point s is now determined from the fill rate condition, i.e.  2KD , μBO  (1 − β) h

20 Stochastic Inventory Models

453

or ⎡∞ ⎤   ∞ 1 ⎣ 2KD 2 2 ⎦ (x − s) fR+L (x)dx− (x − s) fL (x)dx  (1 − β) 2μR h s

(20.2)

s

and subsequently the order-up-to level S from σ 2 + μ2R + S s− R 2μR



2KD h

Note that for an arbitrary demand distribution function, the left part of Eq. (20.2) may not be easy to determine. For a normal distribution function tables exist, see for example, Silver et al. (2017). However, we follow another approach. Because both uR+L and uL are the addition of a number of independent, identically distributed (iid) random variables with known mean and variance, their mean and variance are easily determined. Next, we may fit a mixture of Erlang distributions on the mean and variance or each variable (cf. De Kok 1989; or Tijms 1994) which enables an easy calculation of the left-hand expression of (16.2). It is well known that the class of mixtures of Erlang distributions is dense in the space of all pdf’s, i.e., any pdf can be approximated arbitrarily close by a mixture of Erlang distributions (Schassberger 1973). Finally, we consider the case that the ratio K/h is small or equal to 0. From the initial discussion of (R, s, S)-policies, it immediately follows that s  S, if K  0. Following the same logic as before, we then find that the optimal policy in a periodic review infinite horizon inventory model is of the type (R, S), i.e., at the beginning of each period we bring the inventory position up to level S. This means that the stock used to fulfill the previous period demand is replenished. Indeed, (R, S)-systems are often used when a large number of different SKU’s is ordered periodically (e.g., every week) from the same supplier or by using one logistics service provider. If a periodic delivery schedule for a large number of different SKU’s has been predetermined, then the main ordering costs are fixed throughout the year and hence we may ignore them in the optimization procedure. If R is predetermined in an (R, S) system (which is usually the case) then we are left with a one parameter problem, i.e., the determination of S. Let us take a service perspective again, i.e., we wish to minimize inventory holding costs, subject to a fill rate constraint. Note that an order placed at time t will arrive at time t + L, and any order issued later will not arrive earlier than at time t + L + R. That means, the inventory position after ordering at time t should be sufficient to cover the cumulative demand in the time interval [t, t + L + R). Therefore, the expected on-hand inventory at t + L + R (i.e., just before arrival of the next order) equals S μOH 

(S − x)fL+R (x)dx 0

454

H. Zijm

Similarly, the expected backorder position at time t + L + R, before the arrival of the next order, is ∞ μBO 

(x − S)fL+R (x)dx S

Therefore, noting that on average we order an amount μR at every review instance, to satisfy a target fill rate β, S should be chosen such that ∞ (x − S)fL+R (x)dx  (1 − β)μR S

In the special case of (R, S) inventory systems, we can find an explicit relation between cost and service models. Assume, as before, that inventories are charged at a rate h  rc per item per period, while a shortage is penalized at a rate p per item per period (all costs are charged at the end of a period, just before a new order arrives). Note that, if the net stock at the beginning of interval t + L (t being a review time instant) equals y (just after arrival of an order placed at time t) then the expected costs just before the next review moment t + L + R are equal to H (y) (see the Newsboy equation discussed in the beginning of Sect. 20.3). Hence, if at time t we issue an order such as to return the inventory position back to S, then the expected costs as a result of that action at the end of period t + L + R are ∞ H (S − u)fL (u)du

C(S)  0

S

∞ (S − u)fL+R (u)du + p

h 0

(u − S)fL+R (u)du S

where the last equality follows from taking convolutions of fR and fL . Putting the derivative of C(S) equal to zero yields the following. hFL+R (S) − p(1 − FL+R (S))  0 or FL+R (S) 

p h+p

Thus, knowing the distribution of demand in a time-interval of length R + L, we can explicitly derive the optimal order-up-to level S as a function of p, and also the corresponding fill rate β, via

20 Stochastic Inventory Models

1 β 1− μR

455

∞ (x − S)fL+R (x)dx S

It is intuitively clear, and it can be proven rigorously, that the fill rate determined in this way is a monotone increasing function of p, and vice versa. Hence, by applying a simple bisection procedure, it is easy to determine the value of p (and S) that corresponds to a target fill rate. In the next section, we will discuss multi-stage or multi-echelon inventory systems, in which at each stage n an (R, Sn )-policy is applied, again under a pure cost framework. The determination of the fill rate at the most downstream stage (i.e., the stage facing independent market demand) appears to be a bit more tricky but builds on the cost analysis of a single-stage (R, S) system.

20.4 Multi-stage, Periodic Review Inventory Systems (State-of-the-Art) In this section, we discuss the so-called multi-stage or multi-echelon systems. In the first part, we concentrate on strictly linear systems, consisting of N stages where each stage receives goods from its predecessor and delivers products to its successor. The most upstream stage orders materials from an external supplier and the most downstream stage faces customer demand, see Fig. 20.2. One may think of a chain consisting of a production facility, followed by a central warehouse, followed by a retail shop. Inventory holding costs are incurred at each stage where the final stage may incur additional penalty costs. For example, due to an out-of-stock situation, customer demand cannot be satisfied immediately and must be backlogged. After analyzing serial systems, we turn to 2-stage distribution systems, in which a central depot supplies several local warehouses.

Fig. 20.2 A linear multi-stage inventory system

456

H. Zijm

20.4.1 Linear Multi-stage Inventory Systems Consider first a two-stage linear inventory system, consisting of two stages, denoted by I2 and I1 (following the flow of goods). I2 can order material from an external supplier (with unlimited stock), which is received after a lead time L2 , while I1 orders material from I2 , which is received after a lead time L1 , assuming all materials are available at I2 when ordered. If I2 is short of materials, it delivers as much as possible while the remaining goods are backordered. I2 charges an inventory holding cost per item per period equal to h2 , for items stored at the installation and still in transit to I1 . Items available at I1 are charged at an inventory holding cost of h2 + h1 per item per period while a penalty p per item per period is charged in case of a shortage at stage 1. The additional holding costs h1 in stage 1 may reflect the value added for items in transit between I2 and I1 , e.g., when the transition from I2 to I1 represents a production phase, whereas the upstream transition to I2 is the materials supply from an external source. Note that a backlog at I2 is not charged (at least not directly). Now define the net echelon stock of I1 simply as its net stock, and the net echelon stock of I2 as the stock at stage 2 plus all items in transit from I2 to I1 as well as the net stock of I1 . Hence, the net echelon stock of I2 includes the net (echelon) stock of I1 . Note that the net echelon stock of I2 can be negative although there may still be items in transit to I1 . The echelon inventory position of each stage is defined as that stage’s net stock plus all products ordered but not yet received. Note that the inventory position of I1 does not simply include all items on their way from I2 , but possibly also items still on their way to I2 (i.e., when I2 faces a backlog), in which case I1 may experience a lead time larger than L1 for these delayed items. The concept of echelon stock has been introduced by Clark (1958), after which Clark and Scarf (1960) were the first to show that in a multi-stage system with no fixed ordering costs, an optimal inventory control policy is of the type (S1 , S2 ), meaning that at the beginning of each period each stage In brings its echelon inventory position up to Sn n  1, 2, in an infinite horizon discounted cost framework. The average cost analysis presented below is due to Langenhoff and Zijm (1990) while the computational analysis based upon this framework stems from Van Houtum and Zijm (1991). Define v1 and v2 as the net echelon stock of I1 and I2 , respectively, hence v1 ≤ v2 . Consider the following cases: a. v1 ≥ 0. Because v2 ≥ v1 , the stock at I2 plus in transit between I2 and I1 equals v2 − v1 . Then the total costs summed over the two stages are equal to (h1 + h2 )v1 + h2 (v2 − v1 )  (h1 + h2 )v1 − h2 v1 + h2 v2 b. v1 < 0. Then I1 faces a backlog while the physical stock at I2 plus in transit between I2 and I1 equals v2 − v1 (note that v2 may be both positive or negative). The total costs summed of the two stages are therefore equal to

20 Stochastic Inventory Models

457

h2 (v2 − v1 ) + p(−v1 )  p(−v1 ) − h2 v1 + h2 v2 From the above analysis, we conclude that it is natural to attribute costs h2 v2 to the echelon stock v2 , independent of its sign, while we attribute either (h1 + h2 )v1 − h2 v1 (if v1 ≥ 0) or p(−v1 ) − h2 v1 (if v1 < 0) to x1 . Note that the second term in the costs attributed to v1 is again independent of its sign. If the net echelon stock of I1 and I2 is increased to y1 and y2 at the beginning of a period respectively, then the expected costs at the end of the period are equal to H1 (y1 ) + H2 (y2 ), where y1

∞ (y1 − u)fR (u)du + p

H1 (y1 )  (h1 + h2 )

(u − y1 )fR (u)du y1

0

∞ − h2

(y1 − u)fR (u)du 0

∞  h1

(y1 − u)fR (u)du 0

∞ (u − y1 )fR (u)du

+ (p + h1 + h2 ) y1

∞ (y2 − u)fR (u)du

H2 (y2 )  h2 0

Finally, suppose that at some review time t, an order is placed by I2 so as to increase its echelon inventory position to Y2 and, at time t + L2 , an order is placed by I2 so as to increase its echelon inventory position to Y1 . Note that if there is a shortage of items in stage I2 ) at time t + L2 , I2 may not be able to ship the requested amount to I1 , in which case it faces a backlog. Define, similar as in the preceding section, ∞ Hn (Yn − u)fLn (u)du,

Cn (Yn ) 

n  1, 2

0

It is now easily verified that the expected costs at the end of period t + L2 + L1 are equal to ∞ C2 (Y1 , Y2 )  C1 (Y1 ) + C2 (Y2 ) + Y2 −Y1

(C1 (Y2 − u) − C1 (Y1 ))fL2 (u)du

458

H. Zijm

where the last term reflects the situation that at time t + L2 installation I2 has a net echelon stock of Y2 − uL2 < Y1 (where uL2 denotes the demand in the interval [t, t + L2 )), hence I2 is unable to raise the inventory position of I1 to Y1 . Therefore I1 will only have a net stock position of Y2 − uL2 just after arrival of the (partial) order from I2 at the beginning of period t + L2 + L1 . That is, the last term reflects some indirect penalty cost for I2 . Langenhoff and Zijm (1990) show that the parameters (S1 , S2 ) that optimize the overall cost function can be derived sequentially by first minimizing C1 (Y1 ), yielding S1 , and subsequently C2 (S1 , Y2 ), yielding S2 . This decomposition result was initially proved by Clark and Scarf (1960) in a discounted cost framework and significantly reduces the computation of the optimal base stock policies in a two-stage system. The results above can be extended to an N -stage linear system (with inventory holding costs hn and lead times Ln , n  1, 2, . . . , N , and penalty costs p for the most downstream stage) as follows. Define   ∞ ∞ N  hn H1 (y1 )  h1 (y1 − u)fR (u)du + p + (u − y1 )fR (u)du n1

0

y1

∞ (yn − u)fR (u)du,

Hn (yn )  hn

n  2, . . . N

0

Next, define ∞ Hn (Yn − u)fLn (u)du,

Cn (Yn ) 

n  1, 2, . . . , N

0

and recursively Cn (Y1 , Y2 , . . . , Yn )  Cn (Yn ) + Cn−1 (Y1 , Y2 , . . . , Yn−1 ) ∞ + (Cn−1 (Y1 , Y2 , . . . , Yn−2 , Yn − u)) − Cn−1 (Y1 , Y2 , . . . , Yn−1 ))fLn (u)du Yn −Yn−1

The optimal control policy for the N -stage serial inventory system is a basestock policy characterized by parameters S1 , S2 , . . . , SN . These parameters can be determined sequentially by minimizing C1 (Y1 ), C2 (S1 , Y2 ), and so forth … up to CN (S1 , S2 , . . . , SN −1 , YN ). The importance of the concept of net echelon stock and echelon inventory position (Clark 1958) can hardly be overestimated. The fact that the echelon stock covers all inventories for each stage, downstream up to delivery to the market, implies that stages using echelon stock based control policies always base their decision on the

20 Stochastic Inventory Models

459

forecasted or actual market demand, and not on the demand of the next downstream stage. It is this phenomenon that prevents the amplification of stock variation (as a reaction on demand variation) in a multi-stage system, initially discussed by Forrester (1961) and later analyzed as the so-called Bullwhip effect by e.g., Lee et al. (1997). Naturally, under a centralized control policy, prevention of the Bullwhip effect is easier than in a multi-stage system involving multiple companies or stakeholders. In the latter case, contracts between these companies that include penalties in case of delivery failures should serve to prevent temporary shortages, see e.g., Cachon and Zipkin (1999) or Zijm and Timmer (2008).

20.4.2 Two-Stage Distribution Systems Next, we turn to two-stage distribution models in which one central depot regularly ships products to a number of local warehouses, while stock in this central depot is replenished by an external supplier who is always able to deliver. As before, both the depot and each local warehouse places an order each review time such that its echelon inventory position is brought back to a target level (i.e., stages apply an order-up-to level policy). Any demand experienced at any local warehouse as well as at the depot that cannot be fulfilled immediately is backlogged. What makes the control problem in a distribution system different from that of a serial system is the fact that an allocation decision must be made in case of a (temporary) shortage of depot stock. That is, if the sum of the demands of the local warehouses exceeds the available stock in the central depot at any review and ordering instant, we must decide how much to ship to each local warehouse while backlogging the remaining demand. Below, we formally define the two-stage distribution model, cf. Fig. 20.3. The upstream stage (the central depot) is indexed by N + 1, while the downstream stages (the local warehouses) are numbered 1, 2, . . . , N . All stages order at the same time, i.e., at the beginning of a period of length R. Lead times are denoted by LN +1 (from external supplier to central depot) and L1 , L2 , . . . , LN (from central depot to the local warehouses), inventory holding costs are hN +1 at the central depot and hN +1 + hn for all stock in transfer to or available at local warehouse n, n  1, . . . , N . Each local warehouse n incurs a penalty cost pn per unit per period in the event of a shortage. All costs are incurred based on the net stock at the end of a period. The periodic demand experienced by local warehouse n is denoted by its distribution function FR(n) (with density function fR(n) ) and hence the cumulative periodic customer demand is determined by the convolution FR  FR(1) ∗ FR(2) ∗ · · · ∗ FR(N ) . The cost functions for the local warehouses are derived similarly to those for the final stage in a serial system. Let yn denote the net stock of local warehouse n(n  1, . . . , N ) and let yN +1 denote the net echelon stock of the central depot, all at the beginning of a period of length R, just after arrival of possible orders. Because costs are incurred at the end of a period, we can define, as before,

460

H. Zijm

Fig. 20.3 A two-stage distribution inventory model

∞ Hn (yn )  hn

(yn − u)fR(n) (u)du+

0

∞ (pn + hn + hN +1 )

(u − yn )fR(n) (u)du

n  1, 2, . . . , N

xn

∞ (yN +1 − u)fR (u)du

HN +1 (yN +1 )  hN +1 0

(recall that the demand density fR is the convolution of the individual local warehouse demand densities fR(n) , and hence represents the cumulative downstream periodic demand that equals the decrease of the net inventory position of the central depot in a period of length R). Now, suppose that at some review time t, the central depot manager orders an amount of goods from the external supplier and thereby brings the echelon inventory position of the central depot up to level YN +1 . The order arrives at time t +LN +1 , after which the net echelon stock of the central depot equals YN +1 −uLN +1 where uLN +1 is the cumulative customer demand experienced between t and t + LN +1 . Recall that the net echelon stock of a stage includes all inventories downstream as well. Because costs are incurred at the end of the next period of length R we define, as before

20 Stochastic Inventory Models

461

∞ HN +1 (yN +1 − u)fLN +1 (u)du

CN +1 (YN +1 )  0

(where fLN +1 is the convolution of the individual local warehouse demand densities in a period of length LN +1 ). Next, let the managers of the local warehouses each place an order to raise their inventory position to Yn , n  1, . . . , N , at time t + LN +1 . Now, if YN +1 − uLN +1 ≥

N 

Yn

n1

then all local warehouse orders placed at time t + LN +1 can be fulfilled by the central depot manager, and the remaining stock is held at the central depot for at least one more period. If, on the other hand YN +1 − uLN +1 <

N 

Yn

n1

then a decision has to be made on how much products to send to each local warehouse, i.e., how much stock to allocate to each local warehouse. In other words: what is a good allocation policy in case the central depot temporarily falls short? How to  select order-up-to-levels zn  zn YN +1 − uLN +1 such that N    zn YN +1 − uLN +1  YN +1 − uLN +1 n1

To answer that question, we need to determine the cost effects of any partial order fulfillment zn of local warehouse n. Note that, if at time t + LN +1 the manager of local warehouse n increases its inventory position to zn , the net stock of that warehouse at time t + LN +1 + Ln equals zn − uL(n)n , where uL(n)n denotes the demand experienced by warehouse n in a time period of length Ln . Therefore, the costs experienced by local warehouse n at time t + LN +1 + Ln + R equal ∞ Cn (zn ) 

Hn (zn − u)fL(n) (u)du, n

n  1, 2, . . . , N ,

0

Hence, it seems reasonable to apply a simple myopic allocation policy (MYAL), by solving the following non-linear optimization problem min

z1 ,...,zN

N  n1

Cn (zn )

462

H. Zijm

subject to N 

zn  vN +1

n1

zn ≥ wn , n  1, 2, . . . , N where vN +1 denotes the available net echelon stock of the central depot and wn is the inventory position of local warehouse n just before ordering. Because the functions Cn (yn ) are convex, this non-linear optimization problem is easily solved, see e.g., Langenhoff and Zijm (1990). The problem however is that the optimal solution and therefore also the joint optimal costs are functions of not positions just only vN +1 but also of wn , n  1, 2, . . . , N , i.e., of the specific inventory  before ordering at the local warehouses and not just of their sum Nn1 wn . Therefore, a decomposition result similar to the one for serial systems is not straightforward and indeed Clark and Scarf (1960) already observed that the optimal policies in a two-stage distribution network are not necessarily base stock policies. However, by making a simple additional assumption, a decomposition result similar to the one for serial systems is still within reach. The second constraint in the above formulated allocation problem states that we cannot lower the net inventory position of any local warehouse. Without that constraint the following scenario could occur. In the case of a serious temporary shortage of stock at the central depot, a cost-optimal solution is only possible if the net inventory position at some local warehouses is reduced to help other warehouses to bring their stock to an acceptable level, by applying lateral transshipments. If we do not allow lateral transshipments and still want to skip the final inequalities in the constraint set, we must assume that a cost-optimal solution can be reached without lowering any local warehouse’s net inventory position. This is called the balance assumption (cf. Eppen and Schrage 1981), i.e., we assume that a very odd distribution of inventories over the various local warehouses never occurs and, although the central depot stock may not be sufficient to raise each local warehouse’s inventory position to its target base stock level, the cost-optimal solution allows each local warehouse to at least not reduce its inventory position (and for most of them, to increase it). Under the balance assumption we may apply a Relaxed Myopic Allocation (REMYAL) policy which only depends on vN +1 , as a solution of min

N 

z1 ,...,zN

Cn (zn )

n1

subject to N  n1

zn  vN +1

20 Stochastic Inventory Models

463

The assumption implies that the solution still satisfies zn (vN +1 ) ≥ wn , n  1, 2, . . . , N , i.e., that negative allocation quantities never occur. Fortunately, it turns out that the balance assumption is not a very serious restriction, cf. Van Donselaar and Wijngaard (1987), Eppen and Schrage (1981), Federgruen and Zipkin (1984), not because imbalance never occurs but because its impact on the optimal costs are almost negligible, see also Do˘gru et al. (2009). The optimal REMYAL policy is easily found by applying a simple Lagrange multiplier technique, based on the convexity of the functions Cn (zn ), from the following set of equations (in which λ is the Lagrange multiplier)   ∂ Cn (zn )  hn − (pn + hn + hN +1 ) 1 − FL(n)  λ, (z ) n +R n ∂yn n  1, 2, . . . , N N 

zn  vN +1

n1

We now have a set of N +1 equations for the N +1 unknown variables z1 , . . . , zN , λ, which is easily solved numerically. If all downstream cost parameters are identical, i.e., if hn  h and pn  p for n  1, 2, . . . , N , then we even find a closed-form expression, i.e. FL(n) (zn )  n +R

p + hN +1 + λ p + hN +1 + h

n  1, 2, . . . , N

which is known as the equal fractile rule, cf. Eppen and Schrage (1981) who in addition assumed equal lead times. Moreover, if the demand distribution functions satisfy a normalization property, i.e., if there exists a distribution function ϕ FL(n) n +R such that   zn − μ(n) Ln +R (n) , n  1, 2, . . . N , FLn +R (zn )  ϕ σL(n) n +R (n) where μ(n) Ln +R and σLn +R denote the mean and variance of demand experienced by local warehouse n in a period of length Ln + R, then it is easy to show that (n) zn  zn (vN +1 )  μ(n) Ln +R + σLn +R

with μˆ 

N

(n) n1 μLn +R

vN +1 − μˆ σˆ

(20.3)

 N  (n) 2 and σˆ  n1 σLn +R . In other words: the allocation

quantity zn (vN +1 ) is a linear function of vN +1 . The normalization property holds for

464

H. Zijm

a number of demand distribution functions, including the normal distribution, and is trivially satisfied if all local warehouses behave identically, i.e., face similar demand distributions and have equal cost and lead time parameters. Under the balance assumption, a decomposition result similar to the one for serial systems can also be proven for distribution systems. Define the following cost function CN +1 (Y1 , Y2 , . . . , YN , YN +1 )  CN +1 (YN +1 ) +

N 

Cn (Yn )

n1

∞ +

N  {Cn (zn (YN +1 − u)) − Cn (Yn )}fLN +1 (u)du

 n1 YN +1 − Nn1 Yn

where the final term denotes the induced penalty costs which are incurred if the central depot cannot immediately satisfy the demand of the N downstream local warehouses,  i.e. YN +1 −uLN +1 < Nn1 Y  n and therefore  has to allocate the available stock according to the allocation rule zn YN +1 − uLN +1 . It can be shown that the optimal control policy for such a distribution system is a base stock policy, which is characterized by order-up-to levels S1 , S2 , . . . , SN , SN +1 . For n  1, . . . , N , the optimal parameters Sn are obtained from minimizing the cost functions Cn (Yn ). Subsequently, SN +1 is obtained from minimizing the cost function CN +1 (S1 , S2 , . . . , SN , YN +1 ). Hence, as before, the multi-dimensional optimization problem can be reduced to a series of one-dimensional (convex) optimization problems. Note that the calculation of the base stock level for the central depot requires the calculation of the allocation function zn (vN +1 ) for n  1, 2, . . . N , and in principle for all possible values of vN +1 . This severely complicates numerical solutions. If zn (vN +1 ) is linear, as for instance in (20.3), it appears to be possible to express CN +1 (Y1 , Y2 , . . . , YN , YN +1 ) as a type of newsvendor function which facilitates the calculation considerably. In the final section, we briefly discuss numerical solution procedures. For a complete analysis, the reader is referred to Van Houtum and Zijm (1991, 1997) and to Van Houtum et al. (1996).

20.5 Computational Procedures, Based on Incomplete Convolutions (State-of-the-Art) When evaluating the cost functions of multi-stage serial or distribution systems, we often encounter the situation that the demand distribution functions are generally unknown and at best characterized by their mean and variance, or alternatively by their mean and coefficient of variation. Even if the distribution functions are known exactly (which is rarely the case in any practical setting), we need numerical approximations for the integral forms in the cost function. To that end, Van Houtum and

20 Stochastic Inventory Models

465

Zijm (1991) propose to approximate the distribution functions by mixtures of Erlang distributions, or to fit mixtures of Erlang distributions on the mean and coefficient of variation. This is a well-known procedure, based on a result of Schassberger (1973) who proved that any arbitrary distribution function can be approximated arbitrarily close by a mixture of Erlang distributions. Accurate fitting procedures are discussed by Tijms (1994). Such mixtures appear to be fairly tractable, even when arising in sometimes complex convolutions as is the case in multi-period, multi-echelon inventory systems. A detailed description of the numerical analysis of multi-echelon systems by means of approximations based on mixtures of Erlang distributions is beyond the scope of this chapter. The reader is referred to Van Houtum and Zijm (1991) for details, including all proofs of results mentioned above.

20.6 Further Reading As mentioned in Sect. 20.1, there are numerous papers and books dealing with mathematical inventory theory. A rather complete overview is presented in Silver et al. (2017), which is an update of former versions by partially the same authors. A deeper analytical treatment of some topics can be found in Axsäter (2000), while Zipkin (2000) presents an excellent overview of the foundations of inventory theory. A more managerial oriented book is Fogarty et al. (1991). The case study of IKEA is based on information from a Master’s Thesis project carried out at Lund University, see Rasmusson and Sunesson (2009). In addition, there exists significant theory on spare parts inventory management, see e.g., Sherbrooke (2004), Muckstadt (2005) and Van Houtum and Kranenburg (2015), and Chap. 22 on Maintenance Service Logistics in this volume. Multi-stage or multi-echelon inventory systems under centralized control were initially studied by Clark and Scarf (1960, 1962) and, in quite a different framework, by Forrester (1961). However, significant progress was made in the 1980s, see e.g. Schwartz (1981), Federgruen and Zipkin (1984) and a large number of followers, see for example, Tayur et al. (1999). The analysis of multi-echelon inventory systems presented in this chapter is based on Langenhoff and Zijm (1990), Van Houtum and Zijm (1991) and follow-up papers. Diks and De Kok (1998, 1999) take a direct service-oriented (instead of a pure cost) approach) and develop several accurate approximation algorithms, see also Van der Heijden (1997), Van der Heijden et al. (1997). Capacitated multi-stage production-inventory systems have also been studied in a queueing framework, see e.g., Buzacott and Shanthikumar (1993). Contract management in multi-stage systems under decentralized control (multiple independent players) have been studied by Cachon and Zipkin (1999), Zijm and Timmer (2008), who apply a framework closely related to the one presented in this chapter, and others.

466

H. Zijm

References Arrow KJ, Karlin S, Scarf H (1958) Studies in the mathematical theory of inventory and production. Stanford University Press, Stanford, California Axsäter S (2000) Inventory control. Kluwer, Dordrecht Buzacott JA, Shanthikumar JG (1993) Stochastic models of manufacturing systems. Prentice-Hall, Englewood Cliffs, NJ Cachon GP, Zipkin PH (1999) Competitive and cooperative inventory policies in a two-stage supply chain. Manag Sci 45:936–953 Clark AJ (1958) A dynamic single-item multi-echelon inventory model, RM-2297. The Rand Corporation, Santa Monica Clark AJ, Scarf H (1960) Optimal policies for a multi-echelon inventory problem. Manag Sci 6:475–490 Clark AJ, Scarf H (1962) Approximate solutions to a simple multi-echelon inventory problem. In: Arrow KJ, Karlin S, Scarf H (eds) Studies in applied probability and management science. Stanford University Press, Stanford CA, pp 88–100 De Kok AG (1989) A moment-iteration method for approximating the waiting-time characteristics of the GI/G/1 queue. Probab Eng Inf Sci 3:273–287 Diks EB, De Kok A (1998) Optimal control of a divergent multi-echelon inventory system. Eur J Oper Res 111(1):75–97 Diks EB, De Kok A (1999) Computational results for the control of a divergent n-echelon inventory system. Int J Prod Econ 59(1):327–336 Do˘gru MK, De Kok A, Van Houtum G (2009) A numerical study on the effect of the balance assumption in one-warehouse multi-retailer inventory systems. Flex Serv Manuf J 21(3–4):114–147 Eppen G, Schrage L (1981) Centralized ordering policies in a multi-warehouse system with lead times and random demand. In: Schwartz LB (ed) Multi-level production/inventory control systems: theory and practice. North-Holland, Amsterdam, pp 51–67 Federgruen A, Zipkin PH (1984a) Approximation of dynamic, multi-location production and inventory problems. Manag Sci 30:69–84 Federgruen A, Zipkin PH (1984b) Computational issues in an infinite-horizon, multi-echelon inventory model. Oper Res 32:818–836 Fogarty DW, Blackstone JH, Hoffmann TR (1991) Production and inventory management, 2nd edn. South-Western Publishing Company, Cincinnati Forrester JW (1961) Industrial dynamics. MIT Press, Cambridge, MA Iglehart DL (1963) Optimality of (s, S) policies in the infinite horizon dynamic inventory problem. Manag Sci 9(2):259–267 Langenhoff LJG, Zijm WHM (1990) An analytical theory of multi-echelon production/distribution systems. Stat Neerl 44(3):149–174 Lee H, Padmanabhan P, Whang S (1997) Information distortion in a supply chain: the bullwhip effect. Manag Sci 43(4):546–558 Muckstadt JA (2005) Analysis and algorithms for service parts supply chains. Springer, New York Rasmusson S, Sunesson B (2009) Coordinated inventory control—a case study on its performance compared to the current system at IKEA. MSc thesis project 1002, Lund University, Department of Industrial Management and Logistics Ross SM (1970) Applied probability models with optimization applications. Holden-Day, San Francisco Scarf H (1960) The optimality of (S,s) policies in the dynamic inventory problem. In: Arrow K, Karlin S, Scarf H (eds) Studies in the mathematical theory of inventory and production. Stanford University Press, pp 201–209 Schassberger R (1973) Warteschlangen. Springer, Berlin Schwartz LB (ed) (1981) Multi-level production/inventory control systems: theory and practice. North-Holland, Amsterdam

20 Stochastic Inventory Models

467

Sherbrooke CC (2004) Optimal inventory modeling of systems: multi-echelon techniques. Kluwer Academic, Boston Silver EA, Pyke DF, Thomas DJ (2017) Inventory and production management in supply chains. CRC Press (Taylor and Francis), Boca Raton Tayur S, Ganeshan R, Magazine M (eds) (1999) Quantitative models for supply chain management. Kluwer, Boston Tijms HC (1994) Stochastic models: an algorithmic approach. Wiley, New York Tijms H, Groenevelt H (1984) Simple approximations for the reorder point in periodic and continuous review (s, S) inventory systems with service level constraints. Eur J Oper Res 17:175–190 Van der Heijden M (1997) Supply rationing in multi-echelon divergent systems. Eur J Oper Res 101(3):532–549 Van der Heijden M, Diks E, De Kok A (1997) Stock allocation in general multi-echelon distribution systems with (r, s) order-up-to-policies. Int J Prod Econ 49(2):157–174 Van Donselaar K, Wijngaard J (1987) Commonality and safety stocks. Eng Costs Prod Econ 12:197–204 Van Houtum GJ, Zijm WHM (1991) Computational approaches for stochastic multi-echelon production-distribution systems. Int J Prod Econ 23:223–237 Van Houtum GJ, Inderfurth K, Zijm WHM (1996) Materials coordination in stochastic multiechelon systems. Eur J Oper Res 95:1–23 Van Houtum GJ, Kranenburg B (2015) Spare parts inventory control under system availability constraints. Springer, New York Van Houtum GJ, Zijm WHM (1997) Incomplete convolutions in production and inventory models. Oper Res Spektrum 19:97–107 Van Houtum GJ, Zijm WHM (2000) On the relation between cost and service models for general inventory systems. Stat Neerl 54(2):127–147 Zheng YS, Federgruen A (1991) Finding optimal (s, S) policies is about as simple as evaluating a single policy. Oper Res 39(4):654–665 Zijm H, Timmer J (2008) Coordination mechanisms for inventory control in three-echelon serial and distribution systems. Ann Oper Res 158(1):161–182 Zipkin PH (2000) Foundations of inventory management. McGraw-Hill Inc

Chapter 21

Transportation Management Wouter van Heeswijk, Martijn Mes and Marco Schutten

Abstract This chapter provides an introduction to transportation management. In Sect. 21.1 we describe the basic elements of transportation network design, comprising the selection of modes, transportation units, loading units, and the timing of transportation. A brief introduction to intermodal transportation is used to illustrate the relevance of transportation management in modern logistics. With Sect. 21.2 we move to the advanced topic of orchestrating transportation via intermodal networks by introducing the concept of synchromodal transportation. Furthermore, it highlights several issues specific to contemporary long-haul transportation and last-mile transportation. Section 21.3 presents various state-of -the-art research trends related to the management of transportation in integrated networks. We discuss applications of multi-criteria analysis, multi-agent simulation, and conclude with research directions for the construction of the Physical Internet.

21.1 Transportation Network Design (Basic) Demand for the transportation of goods arises from the fact that goods are typically not produced and consumed at the same location. Although the essence of transporting goods from one location to another appears to be trivial, identifying the best way to do so out of many possible options is not. This section aims to provide insights into the considerations that play a role in transportation management.

M. Mes (B) · M. Schutten Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, Overijssel, The Netherlands e-mail: [email protected] M. Schutten e-mail: [email protected] W. van Heeswijk Center for Mathematics and Computer Science, Amsterdam, The Netherlands e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_21

469

470

W. van Heeswijk et al.

The responsibility for arranging transportation does not always reside with the same actor. In addition, a single actor may have multiple roles within a supply chain. Typically, the transportation process starts with a receiver that places an order at a shipper resulting from a demand for goods. Based on the received orders, the shipper seeks to transport the goods to one or more receivers. The shipper may retain control of the process itself, transporting goods with its own resources, or hire one or more carriers to transport the goods. However, as shippers may lack the volume or logistics expertise to arrange transportation efficiently, they may also decide to outsource the shipment to a Logistics Service Provider (LSP). An LSP might be a carrier itself (known as a 3rd Party LSP or 3PL), but may also be an intermediate party without any (or with limited) physical transportation resources (known as a 4th Party LSP or 4PL). In general, higher volumes result in relatively lower costs due to economies of scale, allowing to identify better consolidation opportunities (bundling of goods) and to utilize transportation resources more efficiently. Although various parties may be responsible for the organization of transportation, they all share the common objective to reduce the transportation costs while complying with regulations and satisfying the required service levels. In the remainder of this chapter, we therefore refer to a decision maker who tries to optimize the transportation processes. We exemplify the complexity of transportation management with the following running example. Example Case: Intermodal Transportation by a Dutch LSP This case is based on the operations of a leading Dutch 4PL service provider active in the European transportation market. This company provides logistics services (i.e., transportation and warehousing) to its customers and has contracts with multiple carriers that have one or more modalities (e.g., truck, train, and barge) available to transport the goods. In this example, we consider several customers (receivers) located in Italy that ordered certain products from a number of producers located in the Netherlands. These producers (the shippers) all hire the Dutch 4PL to orchestrate their transportation activities. The transportation process starts in the Netherlands with a truck picking up the goods at these shippers, thereby consolidating the goods into a single shipment. However, instead of driving directly to Italy, the truck unloads at a nearby inland port to make use of so-called intermodal transportation (see Sect. 21.1.6). Subsequently, the goods are placed in a container and transported to Germany by a river barge, where the container is transferred to a freight train to Italy. After arrival at the train terminal in Italy, the individual shipments are delivered to the final customers using courier services. See Fig. 21.1 for an illustration. For details on the transportation planning decisions faced by this Dutch 4PL and how to support these decisions, we refer to Mes and Iacob (2016). During the entire process, transportation must comply with legislation (cf. Chap. 7), anticipate possible disruptions, and meet the expected delivery dates of the customers, while preferably also yielding a profit in a highly competitive transportation sector. Achieving these objectives requires an advanced level of transportation management.

21 Transportation Management

First-mile transportation \ pre-haulage \ drayage

471

Long-haul transportation

Last-mile transportation \ end-haulage \ drayage

Fig. 21.1 Intermodal transportation example. Icons made by Freepik from www.flaticon.com

The scope of this chapter comprises the construction of physical routes and the selection and timing of transportation modes (e.g., truck, train, barge), thereby aiming to utilize the transportation capacity of the network as efficiently as possible by exploiting consolidation opportunities. By adopting this scope, we focus primarily on the tactical level of decision-making. The outline of the remainder of this section is as follows. Section 21.1.1 introduces the basis structure of a transportation network. Section 21.1.2 provides some basic terminology for transportation networks. Section 21.1.3 discusses the selection of transportation modes, whereas Sect. 21.1.4 discusses the selection of transportation units (e.g., containers, trailers) and loading units (e.g., pallets, roll containers). In Sect. 21.1.5 we describe the role of timing in transportation management. Section 21.1.6 concludes this section with an assessment of intermodal transportation.

21.1.1 Basic Structure of a Transportation Network A transportation network may be described as a set of available logistics services—including both transportation services and bundling services—that may be used to transport goods from origin to destination. In mathematical form, a transportation network is typically expressed as a graph G  {V, A}, with the set of vertices V representing physical locations and the set of arcs A signifying transportation services. The vertices V may be divided into subsets of origins, transfer hubs, and destinations. Here, origins are the pickup points of goods (e.g., factories or warehouses), transfer hubs are points at which goods may be transferred from one transportation mode to another (e.g., ports or railway depots), and destinations are points at which the goods must be delivered (e.g., retailers or households). The role of transfer hubs is twofold; they may allow to switch transportation units from one mode to another (e.g., lifting a container from a ship onto a train), but also to split and bundle goods within individual containers or trailers. Transportation management may relate to direct transportation, i.e., transportation from origin to destination without transfers, typically using trucks, as well as to transportation that utilizes transfer hubs, typically

472

W. van Heeswijk et al.

to combine truck with barge and train. The focus of this chapter is primarily on the latter. The network vertices are connected by arcs that represent transportation services. Without any design efforts, a transportation network might be modelled as a complete graph in which all vertices are directly connected to each other, corresponding to a network in which goods can be transported following the shortest path between any two locations (typically using trucks). However, transportation over longer distances often requires the use of transfer hubs. Direct transportation between individual customer locations is often a financially unattractive option that may safely be omitted from the graph without sacrificing the solution quality. Therefore, realistic transportation networks are typically represented by incomplete graphs; the decision maker must choose a suitable route that connects origin and destination by selecting a connected path of arcs in such a graph. We can represent certain characteristics of a mode by defining them as arc properties. Such properties may include, e.g., departure times, travel speeds, costs, emissions, and idle load capacity. Similarly, we may allocate additional properties to vertices, such as handling times and storage capacity. An important modelling decision is whether to represent the transportation network by a time-expanded graph or a time-dependent graph. Time-expanded graphs represent time properties by separate arcs or vertices, e.g., each train departure on a given line is represented by a unique arc. Time-dependent graphs are defined only in space and use functions to incorporate time properties, e.g., train departures are then reflected by adding a timetable property to the arc. Transportation management requires graphs to be sufficiently detailed to make informed decisions with respect to allocation and timing, but typically omits high-level details that are required on the operational and real-time levels of decision making, such as congestion information.

21.1.2 Terminology of Transportation Networks Transportation networks can often be divided into various segments or network subsets, often with vastly distinct characteristics. For example, intercontinental shipping services and delivery tours in the city are subject to significantly different planning considerations. Such segments are often characterized by their own terminology and research branch. In this section, we introduce some basic terminology of transportation networks. Long-haul transportation is freight transportation that takes place over a long distance. What exactly is considered ‘long’ depends on the context of the transportation. In our example case, we may consider the non-truck segment from the Netherlands to Italy as the long haul. However, also the transportation between two cities located only a few hundreds of kilometres apart might qualify as long-haul transportation. Typically, when we refer to the long haul, we consider a distance that is sufficiently long to consider transportation modes other than truck; transportation management on the long haul often includes transportation by trains and vessels.

21 Transportation Management

473

Short-haul transportation or drayage is transportation taking place over a short distance, often referring to a segment of a longer transportation move. Again, what exactly qualifies as ‘short’ depends on the context. In intercontinental transportation, the last part of the transportation route might comprise the drayage segment between the port and the customers and producers in the so-called hinterland (the area served by this port, which may still be of a considerable size). In parcel transportation between two cities, it might relate to the final delivery tour within the city center. Drayage operations from the origin to the first transfer hub is referred to as first-mile transportation or pre-haulage. Similarly, drayage operations on the route segments from the last transfer hub in the route to the destination is known as last-mile transportation or end-haulage. When planning routes, it is often not necessary to consider all available transportation services. In our running example, the decision maker would not consider to ship goods from the Netherlands to Italy via, e.g., Norway. Instead, the planning focus would be on a subset of the transportation network that is generally oriented from the Netherlands to Italy. Such subnetworks are known as corridors, which may be defined as sets of (contracted) transportation services that connect two areas with a high volume of transportation flows between them.

21.1.3 Selection of Transportation Modes In the selection of transportation modes, decision makers must take into account various metrics. Aside from the costs of each option, factors such as the service level required by the receiver, environmental goals, and reasons of competitiveness may play roles in determining the most suitable modes. We describe five of the most common forms of transportation (Davidsson et al. 2005). • Pipelines are used to transport continuous physical flows of liquids or gases. Although high capital costs are required to construct a pipeline, the variable costs of pipeline transportation are low. They are therefore suitable for transporting high volumes between fixed locations during longer periods • Waterway transportation is the transportation of goods by a ship via sea or inland waters. It is typically the cheapest transportation option for goods that cannot be transported by pipeline, but suffers from low travel speeds and the inaccessibility of many destinations. The sizes of ships vary greatly, with the largest sea vessels being able to carry well over 10,000 containers

474

W. van Heeswijk et al.

• Railway transportation is executed by train over the railway network. Most cargo trains are able to carry over 200 containers, typically against lower costs than road transportation. In addition, it is significantly faster than waterway transportation. However, trains offer low flexibility, as they can only unload at stations located in the railway network. In addition, as trains share a dedicated infrastructure, railroad transportation is limited by the availability of time slots • Road transportation is conducted via the road network. A variety of road vehicles may be used for this, varying from trailer trucks to cargo bikes. Virtually every location can be visited via the road network. However, the costs of road transportation are relatively high. As many locations are inaccessible by train or ship, road transportation typically comprises at least a part of the route. The high flexibility of road transportation still makes it the most common form of transportation • Air transportation is typically conducted by airplanes (an exception are the helicopters used, e.g., for oil rigs and wind turbines). The high speed of this mode is its chief benefit, yet the costs associated with this mode are high as well. Thus, this mode is typically used only when high service levels (e.g., rapid intercontinental delivery) are required. Cargo transportation by air requires the use of airports, the number of locations that may be reached directly by plane is therefore limited

21.1.4 Selection of Transportation Units and Loading Units Aside from the selection of arcs and modes, the decision maker must also determine which transportation units and loading units to use, taking into account the handling requirements during the transportation (Bekta¸s 2017). Transportation units are the devices used to contain and transport goods, most commonly containers or trailers. Containers are standardized transportation units that can be transported by trains, ships, and trucks; their size is often expressed in TEU (Twenty Feet Equivalent Units, leading to a volume of 38.5 m3 ). Most regular containers measure 2 TEU. Due to the standardization of containers, handling, storage, and transportation can be organized in an efficient manner. By using cranes, full truckloads can be loaded or unloaded with a single lifting- or lowering operation. In

21 Transportation Management

475

addition, containers can be transported efficiently by ships, since they can be stacked. Trailers are transportation units on wheels that can be connected to trucks and can be rolled on and off ships and trains, allowing for a seamless integration. However, unlike containers, they cannot be stacked. Besides these two basic transportation units, it might be necessary to take into account other units or specific enhancements to transportation units. For example, liquids such as gasoline are often transported by dedicated tank units. Food often needs to be cooled during transportation, requiring special containers or trailers with cooling systems. In addition, the (un)loading capacities at hubs and receivers must be taken into account. For example, a small retailer may not have a ramp or forklift at its disposal, in which case the trailer might need to be equipped with a tailboard to unload the goods. On a smaller scale, we distinguish between various loading units. Goods may be transported in many different forms, e.g., in the form of boxes, crates, reels, barrels, or bags. Individual items are often combined in loading units such that they become easier to handle. The most commonly used loading units are pallets. Pallets allow bundling a variety of individual items and, due to their standardized form, can be handled easily by forklift trucks or pallet jacks. Other common loading units are roll containers. Unlike pallets, roll containers can be handled without dedicated equipment, allowing more convenient handling for some users. However, in terms of spatial utilization they are typically less efficient. The choice of loading unit may change during the execution of transportation. For example, a batch of parcels may be shipped on pallets, but broken down into individual parcels before the final delivery to end-consumers takes place. A relevant distinction exists between Full-Truck Load (FTL) and Less-than-Truck Load (LTL) transportation. For FTL transportation, consolidation may take place on the level of the mode, e.g., by placing as many containers on a freight train as possible. On the LTL level, consolidation might take place within transportation units, e.g., by combining two half truckloads into a single trailer at a transfer hub.

21.1.5 Timing of Transportation In addition to the selection of the physical route properties and resources, another important aspect of transportation management is the dispatch timing. The time to dispatch certain goods should be determined for each vertex in the route. Returning to our case, any new order received from Italy may directly be dispatched, resulting in the quickest delivery. However, the decision maker might want to wait with dispatching until, e.g., a container is completely filled with goods destined for Italy, thereby optimally utilizing the transportation units. On a larger scale, the barge sailing from the Netherlands to Italy might want to wait until all its container slots are filled. However, one must also take into account the reliability of the transportation service and the timeliness of deliveries. Therefore, dispatch timing is essentially a trade-off between timeliness, reliability, and efficiency.

476

W. van Heeswijk et al.

With respect to dispatch timing, a key distinction is made between static planning and dynamic planning. In static planning, all shipments are known in advance, allowing to optimally allocate them to resources over time. In dynamic planning, shipments are gradually revealed during the execution of the transportation processes (or alternatively, other time-varying circumstances, such as travel times, forced re-planning over time). Practical settings are typically characterized by dynamic characteristics, although the planning itself may still be performed statically over a given time horizon (e.g., periodically). In a dynamic setting, dispatch decisions must be made without fully knowing how they affect future decisions. Thus, one may decide to ship a container that is only half filled, not knowing that another large shipment request that could fill up the container will arrive the next day. To improve the dispatch timing decisions, decision makers might anticipate future requests, e.g., based on historical data. Another relevant aspect of dispatch timing is the uncertainty embedded in the transportation network, e.g., variable travel times or the possibility of disruptions. Although delaying dispatch as long as possible might aid in identifying new consolidation opportunities, it eliminates the flexibility to anticipate disruptions, which may lead to due date violations. In case of transportation with transfers, delaying shipments upstream might reduce consolidation opportunities at the downstream transfer points.

21.1.6 Intermodal Transportation In the previous sections, we have discussed the selection of modes, transportation units and loading units, as well as the dispatch timing. In transportation management, these decisions are often integrally made, attempting to maximize efficiency while meeting targets for timeliness and reliability, and complying with other constraints (e.g., legislation and environmental goals) that may play a role. Such decisions are particularly complex when considering flexibility in modes, dispatch times, and transfer options. Various definitions exist on the concept of intermodal transportation and the closely related terms multi-modal transportation and co-modal transportation, see, e.g., Crainic and Laporte (1997), Janic (2007), and SteadieSeifi et al. (2014). In this chapter, we rather broadly define intermodal transportation as the practice of using one or more transportation modes—with modes that may be of the same type or different—to transport containerized goods from origin to destination, using transfer hubs to connect the modes and facilitate consolidation of goods. Examples of transportation forms falling under this definition are (i) goods transported by two distinct trucks, swapping the container at a transfer hub where additional goods are loaded in the container, (ii) a container transported partially by train and partially by ship, and (iii) goods delivered by a single truck while having other realistic transportation options available as well.

21 Transportation Management

477

By considering multiple modes, one may take advantage of the benefits of a specific transportation mode. For example, a large barge can hold hundreds of containers; clearly sending a single barge from the Netherlands to Italy is more efficient than hundreds of trucks. From the last transfer hub onwards, we can benefit from the flexibility of trucks to reach all customer locations. Another benefit of using multiple modes is that it may facilitate the bundling of goods for only a part of the route, e.g., goods from the Netherlands destined for Germany may be combined with goods destined for Italy on a part of the route, as they are transported in the same direction. Although intermodal transportation has distinct benefits, it also introduces several challenges. Switching modes requires physical transfers of transportation units or loading units (e.g., at a port or consolidation centre), introducing additional costs to the transportation process. To achieve a financial benefit, these costs should be compensated by a higher transportation efficiency. The shorter the distance between origin and destination, the more challenging it is to compete financially with direct transportation (Janic 2007). Intermodal transportation also introduces various planning challenges. The first is the mode selection. As discussed before, each mode has its own benefits and drawbacks, with the primary trade-off being between costs and service level. Furthermore, transportation units and loading units should be chosen such that they allow for efficient handling between modes. Finally, at each vertex in a route multiple dispatch timing decisions might be made. The combination of all these options may result in a high number of possible schedules. In such complex environments, high-quality transportation management is essential to meet the high financial, environmental, and service standards that are commonplace in contemporary logistics. In practice, the complexity of intermodal transportation is partially mitigated by fixing various options in transportation contracts. For example, a shipper may agree a fixed price to ship a given number of containers per train every month, or agree to use a specific transportation mode. As a result, traditional intermodal transportation is mainly concerned with decisions on the strategic and tactical levels. Although contracting reduces uncertainty in the network and simplifies planning, it also results in sub-optimal use of transportation networks and ignores operational circumstances that might prompt alternative decisions. In Sect. 21.2, we define synchromodal transportation, which focuses on the flexible use of intermodal networks and therefore prompts a shift towards operational and real-time decision-making.

21.2 Design of Integrated Transportation Networks (Advanced) As discussed in Sect. 21.1, intermodal networks embed the potential to manage transportation more efficiently than direct transportation allows. Although intermodal transportation is far from a new phenomenon, optimizing the use of intermodal networks might well become a necessity due to recent trends in logistics. Devel-

478

W. van Heeswijk et al.

opments such as e-commerce and lean storage management result in smaller, more fragmented, and more frequent shipments, which—especially when coupled with higher service standards—make it more difficult to efficiently utilize transportation resources. Many individual decision makers have reached the limits of what is achievable by optimizing their internal processes; to make improving steps, cooperation with other actors is of vital importance. By sharing resources, information, and shipments, higher efficiency gains might be attained than achievable by individual actors. To facilitate such gains, the establishment of transfer hubs and efficient operations for (un)bundling goods are essential. The increasing shift from direct transportation towards intermodal transportation also has a notable impact on the scope of transportation management. Aside from the financial motivations to improve transportation efficiency, the negative effects that freight transportation has on the environment and the society are becoming increasingly relevant. The fragmentation of freight flows, growth of the overall population, and higher living standards are key drivers behind the growing number of transportation movements, amplifying the negative side effects of transportation. From an environmental perspective, emissions are the chief hazard of freight transportation. Greenhouse gases are known to contribute to climate change, whereas other emissions have polluting effects on the environment. Important societal effects are the negative contribution of transportation to congestion, the effects of emissions on health, noise hindrance, and reduced traffic safety. New legislation is continuously being developed to encourage or enforce decisions that take into account environmental aspects; also, companies themselves become more concerned with an environmentally sustainable business and corresponding image. Thus, transportation decisions are no longer made from a purely financial perspective. A decision maker might for example prefer barge transportation to road transportation due to the lower emissions per container. Section 21.2.1 discusses how the concept of synchromodal transportation is used to utilize intermodal networks more efficiently. Section 21.2.2 focuses explicitly on long-haul transportation in modern logistics, whereas Sect. 21.2.3 focuses on lastmile transportation.

21.2.1 Synchromodal Transportation The term ‘synchromodality’ is relatively new in the world of transportation. Synchromodal transportation might be viewed as an extension of intermodal transportation in which the aim is to select the best combination of modes for each individual shipment. Unlike traditional intermodal transportation, synchromodal transportation is also not restricted to containerized transportation only. Synchromodal transportation focuses explicitly on seamless connections between modes and a high degree of flexibility based on the prevailing circumstances in the network (Tavasszy et al. 2015). As noted before, traditional intermodal transportation typically involves contractually fixed transportation flows. Although this reduces the uncertainty in the network and

21 Transportation Management

479

simplifies planning, it also results in sub-optimal use of the transportation network and ignores operational circumstances that might prompt alternative decisions. As synchromodality implies a large degree of freedom in selecting the best solution for each individual shipment, this concept is challenging from a transportation management perspective. We divide our assessment of managing synchromodal networks into three categories: physical management, information management, and financial management. With respect to physical management, we may distinguish between offline planning and online planning. Offline planning is concerned with the construction of a transportation network with sufficient flexibility to facilitate synchromodal transportation. Typically, a single party does not own a transportation network of the required magnitude, such that a suitable network usually consists of own transportation modes and transfer hubs as well as subcontracted transportation resources (Van Riessen et al. 2015). From a transportation management perspective, the key challenge is to construct a network that is dense both geographically and with respect to timing, such that the decision maker can truly take advantage of planning flexibility. However, the complexity of managing the network also increases with its temporal-spatial density. The construction of transportation networks over space and time is known as Service Network Design; we refer to Crainic (2000) for a detailed description. Online planning focuses on the actual routing of individual orders transported via the synchromodal network. As each order is individually planned based on the prevailing state of the network, traditional manual planning is often too timeconsuming for synchromodal planning. Efficient algorithms are indispensable to quickly identify good routes and respond to disturbances. To this end, Bock (2010) presents a method based on local search principles, explicitly focusing on planning in a real-time environment. The ability to respond to disruptions during the execution of routes clearly contrasts with traditional intermodal transportation. Another synchromodal planning algorithm is presented by Mes and Iacob (2016). They describe a constructive algorithm that efficiently plans synchromodal transportation, proposing a solution method based on the k-shortest path problem. This approach may be used to present multiple high-quality solutions to a human planner, allowing the planner to address possible considerations not reflected in the mathematical model, without having to deal with large numbers of solutions manually. The second category to be discussed is information management. To offer the desired flexibility of synchromodal transportation, decision makers must be able to respond rapidly to changes in the network. Therefore, the real-time exchange of timely and accurate information is essential. Examples of information relevant for transportation management are the idle capacity of modes, order properties, delays in travel time, and disruptions at transfer hubs. Singh (2014) describes an outline for an IT platform facilitating synchromodal transportation, stating that much work still needs to be done to achieve such a generic platform. The third and final category that we address here is the financial perspective. Traditionally, transportation is paid per utilized transportation mode. In synchromodal transportation, modes are not selected in advance, yet it is eminent that potential additional costs for the sake of system-wide efficiency (e.g., a detour to warrant higher

480

W. van Heeswijk et al.

container utilization) should not be charged to individual shippers. In addition, many shippers are cautious or reluctant to hand over control of the transportation process, preferring to have a high degree of autonomy in arranging their transportation (Van Riessen et al. 2015). To overcome these barriers, shippers must be provided with the proper incentives to consider synchromodal transportation. Revenue management based on desired service levels is already common in sectors such as the airline industry. In freight transportation, comparable pricing mechanisms are less widespread.

21.2.2 Long-Haul Transportation In Sect. 21.1, we pointed out that there is no fixed definition of the distance that constitutes long-haul transportation. In the context of intermodal transportation, we might argue that the distance should be sufficiently long to consider reasonable alternatives to direct transportation. The minimum distance for intermodal transportation to be financially competitive with direct transportation is believed to be a distance somewhere between 300 and 500 km (Wagener 2014). Developments in automated handling and information exchange might cut costs in the future and thereby reduce the minimum required distance. When viewed from a holistic perspective, managing long-haul transportation is not only concerned with routing decisions. Aspects that also must be managed are, e.g., the positioning and distribution of empty containers, the assignment of crews to modes, fuel management, and the frequency of long-haul services (Crainic 2003). Such decisions should be made integrally to create a balanced long-haul transportation network that is aligned with the expected transportation flows. While fleet management is already challenging for individual carriers, the challenges become even more pronounced when considering cooperation with subcontractors. Coordination, pricing, and risk control are essential factors in integrated fleet management. An aspect of long-haul transportation that we highlight here is the influence of regulation on transportation management. On the one hand, liberation of markets, globalization, and alignment of national regulations have made it easier to arrange cross-border transportation. On the other hand, regulations with respect to safety and the environment have become more stringent. Examples of regulations affecting transportation management are break schedules and rest periods of drivers, the availability of track capacity for railroad transportation, and the restrictions of maritime law on vessel operations. These topics exemplify the complicating effects that regulation has on transportation management. We conclude this section with an example of transportation management in longhaul transportation. Van Heeswijk et al. (2016) study the dynamic selection of longhaul services in intermodal networks (trucks, barges, and trains) for LTL goods, and present a methodology to support this decision. For each incoming transportation request, the proposed methodology picks the k best paths out of a large number of potential solutions and subsequently checks for possible consolidation opportunities

21 Transportation Management

481

within containers. Routes may be replanned (if not already started) to allow the consolidation of goods and to cut the overall transportation costs, taking advantage of the bundling capacity of transfer hubs. Aside from financial concerns, mode speeds and emissions per route may play a role in the decision. This example illustrates how the variety of transportation modes and the presence of transfer hubs can be utilized for more cost-efficient and environment-friendly long-haul transportation.

21.2.3 Last-Mile Transportation In this section, we discuss transportation management considerations in last-mile transportation, with a focus on last-mile transportation in urban environments. Although we only discuss last-mile transportation here, similar considerations play a role in first-mile transportation. Last-mile transportation covers only a small part of a transportation route, yet it accounts for a disproportionally large part of the total transportation costs (Gevaers et al. 2011) and poses significant challenges. Due to slow and congested traffic, as well as many pickup and delivery stops that require handling operations, lastmile transportation is time-consuming and fuel-inefficient (e.g., due to bridges with weight limits and access time constraints). In addition, environmental and societal concerns weigh heavier on the last mile. Various emissions—such as SOX , NOX , and particulate matter—have strong local effects on health and the environment that become more pronounced in densely populated urban areas. Societal concerns include safety, noise hindrance, and congestion. Therefore, there is increased pressure to arrange last-mile transportation in a way that is both efficient and environmentfriendly. Heavy trucks are efficient modes for transportation over longer distances, but in urban environments, where last-mile transportation typically takes place, their drawbacks become severe (Dablanc 2007). Therefore, we might prefer to deploy smaller, more environment-friendly vehicles, such as electric vans for deliveries within the city. Sometimes, such a change of transportation mode might even be enforced, e.g., due to a ban on certain trucks within the city center (e.g., trucks that do not meet certain engine requirements). Thus, we might want to decouple the freight flows stemming from the long haul at a transfer hub, which has the additional advantage that goods from different inbound trucks may be bundled, allowing for more efficient last-mile delivery routes. A common solution to reduce the number of heavy trucks in urban areas is the concept of urban consolidation centers. These transfer hubs are typically located at the edge of urban areas, where inbound trucks may unload the goods destined for the city. At the hub, goods are subsequently bundled and delivered with smaller vehicles such as delivery vans or electric cargo bikes. The planning problem faced by the transfer hub is twofold, as it must (i) plan routes in a complex environment with congestion, delivery windows, and possible access constraints and (ii) decide on the dispatch timing, anticipating future consolidation opportunities. An algorithm

482

W. van Heeswijk et al.

for the congestion routing problem with delivery windows is presented by Kok et al. (2012). This model takes into account the variation in travel speeds as a function of the time of the day, allowing to create congestion-avoiding routes. As a result, service levels may be increased and costs may be decreased. The dispatch timing problem is addressed in Van Heeswijk et al. (2017b). In this study, goods arrive at the transfer hub from the long haul. The hub operator seeks to bundle and deliver these goods as efficiently as possible, but does not know in advance, which goods arrive or when exactly they arrive. Thus, in the timing of dispatch, the operator has to consider the potential customer locations, delivery windows, and the volume of goods. Based on the expected arrival process, the operator is able to construct dispatching policies that take into account future consolidation opportunities. Dellaert et al. (2016) argue that for larger cities, a single consolidation center does not suffice to arrange last-mile transportation effectively, instead opting for multi-echelon systems that typically contain one layer of consolidation centers and one layer of smaller satellite facilities. Mega-cities may even require more than two layers of transfer hubs. Such structures are challenging to handle from a transportation management perspective, as they require highly integrated decision-making with respect to facility selection, dispatch timing, and mode selection. Besides last-mile transportation in an urban setting, we also briefly describe its use within an intermodal transportation context. Last-mile transportation in the context of intermodal transportation, which we denote by drayage, entails decisions comparable to urban transportation, but may focus on different aspects. An important aspect is the time it takes to load or unload a container at a customer. Decoupling the transportation unit makes it possible for trucks to move on while such operations take place (Pérez Rivera and Mes 2017). For example, a truck might leave an empty trailer at a customer and pick up a full trailer at the same customer, leaving the empty trailer to be filled and to be picked up later on by another truck. Also the distinction between containers and trailers is important; trucks should be equipped with the proper chassis to pick up a container, whereas trailers have their own chassis. Additional considerations arise when considering rigid trucks (having a transportation unit fixed to the frame) pulling a trailer, or trucks transporting two containers; in those cases decoupling decisions must be made for multiple transportation units per vehicle.

21.3 Current Research Trends in Integrated Transportation Networks (State-of-the-Art) Throughout this chapter, we have touched upon various developments that are relevant to the future of transportation management. We briefly reiterate these developments before linking them to various research trends in transportation management. We have asserted that cooperation and coordination between actors in logistics is becoming increasingly important, due to globalization, higher service level standards, and fragmentation of freight flows. Furthermore, environmental and societal

21 Transportation Management

483

concerns are gaining prominence in both legislation and decision-making. These developments limit the efficiency that may be achieved by individual actors. Instead of direct transportation of goods from origin to destination, transportation moves are increasingly characterized by the use of multiple modes and transfer hubs, which are often controlled by various actors. Thus, transportation management is no longer concerned only with managing the set of resources controlled by the decision maker, but also the integration with network segments of other actors. Although we have mentioned several challenges that complicate the planning of intermodal transportation, there are also trends that provide opportunities. Important developments in recent years have occurred in the tracking of vehicles, transportation units, and individual shipments, as well as the real-time exchange of information. Many vehicles nowadays are equipped with GPS and a board computer, transferring information about the whereabouts of the vehicle and the progress of its route. RFID chips and barcodes are used to read the properties of goods with no or minimal handling involved. Developments in information systems and the speed of data transfers allow to quickly share this information, enabling for instance the identification of idle space in a container. Furthermore, today’s computing power enables the processing of large amounts of real-time data or to evaluate millions of routes within limited time. Combined with increasingly sophisticated algorithms, planners are able to find high-quality solutions even for large and complex networks. Finally, innovations in automation lower the costs of certain transportation and handling operations, paving the road to more integrated transportation networks. A recent example of automation is the deployment of automated guided vehicles in ports, which collect containers at quayside cranes and transport them to the container stacks without human intervention. Although many trends may be identified in transportation management research, the focus of this section is on topics that explicitly take into account the importance of collaboration, non-financial objectives, and the integration of transportation networks. In Sect. 21.3.1, we describe the concept of multi-criteria analysis. Section 21.3.2 briefly discusses multi-agent systems that may be used to model processes in transportation management. Section 21.3.3 concludes the chapter with an introduction to the Physical Internet, giving an outlook to the expected developments in transportation management in the next decades.

21.3.1 Multi-Criteria Analysis Transportation management is no longer driven purely by financial incentives, but also takes into account aspects such as safety, emissions, and service levels. Deploying decision making tools that focus solely on a single performance indicator therefore no longer suffices for many decisions in contemporary transportation management. Instead, multiple criteria often need to be taken into account to reach satisfactory solutions.

484

W. van Heeswijk et al.

Velasquez and Hester (2013) provide an overview of multi-criteria decision methods, pointing out various methods that are particularly suitable in transportation settings. Assigning weights to criteria, ranking alternative solutions, and handling uncertainty in input data are important properties of good multi-criteria decision methods. Multi-criteria analysis provides a structured approach to make transportation management decisions, taking into account the complexity of the sector and the increasing availability of data. Macharis et al. (2009) extend the notion of multicriteria analysis by also adopting a multi-actor perspective, which may be used for, e.g., evaluating transportation policies and the selection of transportation technologies. From a transportation management perspective, the main benefit of considering multiple actors is that the objectives of stakeholders, such as shippers and residents, are explicitly taken into account, increasing the chances of identifying solutions that receive system-wide support. It also illuminates gaps in the perspectives of different actors. As transportation management is becoming increasingly dependent on cooperation with external parties and satisfying non-financial criteria, it is of paramount importance that all involved actors deem the proposed solution acceptable and are willing to commit time and resources to aid in achieving system-wide objectives. There are multiple uses for the application of multi-criteria analysis in transportation management. For example, in the selection of supply chain partners, nonfinancial criteria, such as reliability, capacity, and connectivity (supported by appropriate information systems) are important aspects to consider, yet these are often difficult to translate in financial terms. Other common non-financial criteria in transportation management are, e.g., emissions and route durations. Also when setting up collaboration structures between various actors (e.g., constructing an integrated transportation network), multi-criteria analysis may be applied to assess how the structure affects each of the involved actors and whether the requirements of each actor are satisfied. Nowadays, large amounts of data are available to aid decision-making in transportation. As modern computers are well-equipped to handle large data sets, multicriteria analysis has become a powerful tool to aid decision making in complex environments. Research efforts are increasingly directed towards algorithmic approaches that handle many criteria and their corresponding data sets. With the ongoing integration of transportation networks and the abundant availability of both data and computing power, multi-criteria analysis is expected to remain a relevant research topic for transportation management in the foreseeable future.

21.3.2 Agent-Based Systems We have emphasized the importance of integrating transportation networks in modern logistics, indicating that multiple actors must cooperate or at least coordinate their activities to benefit from bundling resources, information, and shipments. Often these networks lack a clear power structure between actors, therefore transportation management essentially has to deal with various autonomous decision makers whose

21 Transportation Management

485

support is required to successfully implement solutions. A suitable method for evaluating and optimizing decisions in such an environment is an agent-based system, in the form of a multi-agent system or agent-based simulation. Chapter 27 is dedicated to agent-based systems in logistics; we therefore restrict ourselves to a very basic introduction here. Agent-based systems model the behaviour of real-life actors by pieces of software called agents. The system embeds multiple agents that autonomously make decisions and might be able to communicate and interact with each other. This decentralized structure mimics a real-life setting in which each agent attempts to make decisions that optimize its own objective function, disregarding whether this also contributes to the desired system-wide solution. Depending on the purpose for which the system is developed, the intelligence of agents might range from very basic decision rules to sophisticated planning algorithms. Although the body of transportation research literature contains vast numbers of highly advanced decision methods, most studies focus on optimizing from the perspective of individual or centralized decision makers. As the focus of transportation management increasingly shifts towards integrated transportation networks, additional attention for the application of agent-based systems is required to tackle the problems faced in modern logistics. In transportation taking place over longer distances, the focus of agent-based analysis is often on either horizontal or vertical collaboration in supply chains. Horizontal collaboration is often concerned with the coordination between transportation services offered by different agents and achieving a workload distribution that is both efficient and agreeable to the agents. Vertical collaboration often focuses on information exchange that facilitates flexible planning of freight flows. In first- and last-mile transportation, agent-based simulation is used to evaluate solutions that affect multiple actors within this transport segment. We illustrate the application of agent-based simulation in last-mile transportation using a case study on urban freight transport in the city of Copenhagen. Case Study: Agent-Based Simulation to Evaluate Urban Logistics Schemes This case is based on a study on urban freight transport in the city of Copenhagen. In this study, the stakeholders in urban logistics, i.e., carriers, receivers, an urban consolidation centre, and the municipality, are modelled as autonomous agents. Carriers arrive from the long-haul with goods destined for the city. They may enter the city themselves or outsource the last-mile transportation to the consolidation centre. The receivers—located in the lowemission zone of the city—place transportation orders. Subsequently, they must spend time on receiving each shipment. In addition, they need to fulfil valueadding services such as storage, labelling clothes, or collecting waste. They might outsource the last-mile delivery as well as the value-adding services to the consolidation centre. Both the carriers and the receivers aim to minimize

486

W. van Heeswijk et al.

their costs and must decide whether to outsource to the consolidation centre. The centre aims to maximize its profit. Efficient routing and handling help to achieve this goal, which in turn requires the collaboration with as many users as possible. Finally, the municipality attempts to reduce the environmental and societal impact of freight transportation by imposing access restrictions and subsidy measures. Figure 21.2 illustrates the test setting. To study the interaction between the decision makers, an agent-based simulation framework has been designed. This framework splits decision-making into three levels. On the strategic level, the administrator sets policies such as subsidy measures and access restrictions for trucks; these decisions are fixed for multiple years. Tactical decisions are made for two-month periods. First, the consolidation centre adjusts its price levels based on the handled volume. Subsequently, carriers and receivers decide whether to outsource to the consolidation centre. On the operational level daily routing decisions are made by both carriers and the consolidation centre. By applying agent-based simulation, the interaction between the agents, their choices over time, and the eventual impact on both financial and environmental performance indicators can be monitored. In particular, the most effective combinations of administrative policies to support the use of the consolidation centre can be evaluated. As individual measures typically have insufficient impact, simultaneous implementation of multiple measures is often necessary. For example, combining temporary subsidies to carriers for using the UCC, with access time restrictions for heavy trucks, might be efficient when applied in conjunction, whereas the measures have limited impact when applied in isolation. The simulation framework enables to test many of such combinations within a short amount of time. This case illustrates that transportation management in an urban context often cannot be regarded as an optimization process by a single actor, but should instead be viewed in the light of interaction with other actors that strive to accomplish certain objectives (both financial and non-financial) for themselves. For further details on this study, we refer to Van Heeswijk et al. (2017a).

Due to the applicability of agent-based simulation for evaluating solutions involving multiple autonomous decision makers, this research branch is becoming increasingly relevant in transportation management.

21.3.3 Physical Internet In this chapter, we have discussed the shift of transportation management towards managing integrated networks, stressing the importance of collaboration and sharing

21 Transportation Management

487

Fig. 21.2 Map of the Copenhagen test setting, showing the receiver locations (green) in the shaded low-emission zone, the UCC location (blue marker), and the long-haul exit point for carriers (red marker) (OpenStreetMap.org)

transportation resources to deal with future challenges. The Physical Internet is a logistics concept that stretches this notion to the extent of a completely integrated global logistics system (Montreuil 2011). It essentially labels each individual shipment with an origin, a destination, and certain constraints to be fulfilled, but this framework provides complete freedom on how to transport the shipment from origin to destination. This flexibility allows to extensively take advantage of bundling operations and idle transportation capacity. The name of the concept is derived from its analogy to the digital internet, in which the user is not concerned with how information is transferred precisely, but only with its reliable and timely arrival. More specifically, the Physical Internet relates to IPv4, an internet protocol that frequently fragments and reassembles information packages during the transmission from sender to receiver. For an in-depth analysis of the Physical Internet, we refer to Chap. 31. The large flexibility of the Physical Internet implies that routes are fairly complex. Routes planned in the spirit of the Physical Internet typically consist of multiple modes and are typified by frequent bundling and unbundling at transfer hubs. The

488

W. van Heeswijk et al.

Fig. 21.3 Roadmap to the Physical Internet (European Technology Platform ALICE)

key difference between the Physical Internet and the concept of synchromodal transportation as described in Sect. 21.2, is the absence of a central control mechanism. Although synchromodal transportation also seeks to find the optimal route for each individual order through an intermodal network, it typically relies on a single decision maker, e.g., a 4PL orchestrating the transportation. In contrast, the Physical Internet assumes a decentralized control mechanism, with automated and real-time dispatch decisions taking place during the execution of the routes. Figure 21.3 provides a roadmap to the Physical Internet, presenting the main challenges per decade on various subdomains that should ultimately converge to a truly seamless logistics system. The eventual goal of the system is envisioned to be realized by 2050. Collaboration, integration, and automation are key components of the roadmap. The steps required to develop the Physical Internet present an ambitious view on the future of transportation management.

21 Transportation Management

489

Although a vast amount of research is conducted within the subdomains of the Physical Internet, research on the concept itself is still in its infancy. However, some trends may already be identified. Crucial to the success of the Physical Internet is the availability of smart containers that communicate real-time information and can be transferred easily between modes. Montreuil et al. (2010) describe the required modularity and interconnectivity of such containers to handle various dimensions of goods and to suit different types of handling equipment, varying from quayside cranes to retail equipment. Such modular containers are supposed to reduce handling costs, as they do not have to deal with goods of many different shapes and sizes. Aside from suitable transportation units, the facilitating role of transfer hubs is also essential. Efficient processing of modular containers is important to minimize the time and costs spent on handling operations. On the technological side, real-time information exchange is a major topic. Especially in harsh environments, such as at sea, timely and reliable information exchange might be a considerable trial. Another important challenge that must be solved for the Physical Internet to be successful is its resilience with respect to disruptions, such as congestion or inaccessibility of a hub. Some customers require higher service levels than others, yet planning robust routes is challenging when considering decentralized decision-making. Depending on the willingness to pay, distinct risk profiles may be assigned to differentiate goods, thus creating a trade-off between efficiency and robustness. The categories that we mentioned in Sect. 21.2.1 for research directions on synchromodal transportation (physical, information, financial) are undoubtedly relevant for the Physical Internet as well. Research on the Physical Internet seemingly lacks direct applications, yet the long-term vision is vital to properly direct the research efforts of transportation management in integrated networks. The topics of network integration, decentralized decision-making, real-time information exchange, and non-financial criteria will be prominent research branches in the future of transportation, prompting a drastic break with the traditional view on transportation management.

21.4 Further Reading We conclude this chapter with a number of suggestions for further reading. For Sect. 21.1, we highlight the following works. The book of Bekta¸s (2017) provides a general introduction into freight transportation, delving deeper into the topics covered in this chapter. Crainic (2000) focuses on the mathematical analysis of transportation management and provides a review of modelling efforts in the field of Service Network Design. The literature review of SteadieSeifi et al. (2014) summarizes and classifies planning methods applied in intermodal transportation. We point out four works relevant to Sect. 21.2, covering the topics of long-haul transportation, last-mile transportation, and synchromodal transportation. Crainic (2003) provides an overview of modelling efforts in long-haul freight transportation. A generic model of the two-echelon capacitated routing problem is presented by

490

W. van Heeswijk et al.

Dellaert et al. (2016). Van Heeswijk et al. (2017b) focus on the dispatching problem of a transfer hub and develop an algorithm to solve this problem. Finally, Tavasszy et al. (2015) give an introduction into synchromodal transportation. The Ph.D. thesis of Van Heeswijk (2017) consists of various planning methods for long-haul and last-mile transportation, and an agent-based simulation framework to evaluate urban logistics schemes. As such, it covers multiple topics addressed in this chapter. Macharis et al. (2010) describe the application of multi-actor, multicriteria analysis on transportation problems, while also discussing the theoretical background of the technique. Montreuil et al. (2010) provide an introduction to the Physical Internet and point out various research directions relevant for the future of transportation management.

References Bekta¸s T (2017) Freight transport and distribution: concepts and optimisation models. CRC Press, Boca Raton, United States Bock S (2010) Real-time control of freight forwarder transportation networks by integrating multimodal transport chains. Eur J Oper Res 200(3):733–746 Crainic T (2000) Service network design in freight transportation. Eur J Oper Res 122(2):272–288 Crainic T (2003) Long-haul freight transportation. In: Handbook of transportation science. Springer US, Boston, United States, pp 451–516 Crainic T, Laporte G (1997) Planning models for freight transportation. Eur J Oper Res 97(3):409–438 Dablanc L (2007) Goods transport in large European cities: difficult to organize, difficult to modernize. Transp Res Part A: Policy Pract 41(3):280–285 Davidsson P et al (2005) An analysis of agent-based approaches to transport logistics. Transp Res Part C: Emerg Technol 13(4):255–271 Dellaert N, Saridarq F, Van Woensel T, Crainic T (2016) Branch & price based algorithms for the two-echelon vehicle routing problem with time windows. CIRRELT, Montreal, Canada, pp 1–39 Gevaers R, Van de Voorde E, Vanelslander T (2011) Characteristics and typology of last-mile logistics from an innovation perspective in an urban context. City distribution and urban freight transport: multiple perspectives. Edward Elger, Cheltenham, United Kingdom, pp 56–71 Janic M (2007) Modelling the full costs of an intermodal and road freight transport network. Transp Res Part D: Transp Environ 12(1):33–44 Kok A, Hans E, Schutten JMJ (2012) Vehicle routing under time-dependent travel times: the impact of congestion avoidance. Comput Oper Res 39(5):910–918 Macharis C, Astrid De Witte A, Ampe J (2009) The multi-actor, multi-criteria analysis methodology (MAMCA) for the evaluation of transport projects: theory and practice. J Adv Transp 43(2):183–202 Macharis C, De Witte A, Turcksin L (2010) The multi-actor multi-criteria analysis (MAMCA) application in the Flemish long-term decision making process on mobility and logistics. Transp Policy 17(5):303–311 Mes M, Iacob M (2016) Synchromodal transport planning at a logistics service provider. Logistics and supply chain innovation. Springer International Publishing, Cham, Switzerland, pp 23–36 Montreuil B (2011) Toward a Physical Internet: meeting the global logistics sustainability grand challenge. Logist Res 3(2):71–87

21 Transportation Management

491

Montreuil B, Meller R, Ballot E (2010) Towards a Physical Internet: the impact on logistics facilities and material handling systems design and innovation. In: Gue K (ed) Material handling industry of America. Charlotte, United States, pp 305–327 Pérez Rivera A, Mes M (2017) Scheduling drayage operations in synchromodal transport. International Conference on Computational Logistics. Springer International Publishing, Cham, Switzerland, pp 404–419 Singh P (2014) Developing a service oriented IT platform for synchromodal transportation. Springer, Berlin-Heidelberg, pp 30–36 SteadieSeifi M et al (2014) Multimodal freight transportation planning: a literature review. Eur J Oper Res 233(1):1–15 Tavasszy L, Behdani B, Konings R (2015) Intermodality and synchromodality [Online]. https://pa pers.ssrn.com/sol3/papers.cfm?abstract_id=2592888 Van Heeswijk W (2017) Consolidation and coordination in urban freight transport. PhD dissertation, University of Twente, Enschede, The Netherlands Van Heeswijk W, Larsen R, Larsen A (2017a) An urban consolidation center in the city of Copenhagen: a simulation study. In: Beta working paper series, vol 523 Van Heeswijk W, Mes M, Schutten J (2017b) The delivery dispatching problem with time windows for urban consolidation centers. Transp Sci. https://doi.org/10.1287/trsc.2017.0773 Van Heeswijk W, Mes M, Schutten J, Zijm W (2016) Freight consolidation in intermodal networks with reloads. Flex Serv Manuf J. https://doi.org/10.1007/s10696-016-9259-1 Van Riessen B, Negenborn R, Dekker R (2015) Synchromodal container transportation: an overview of current topics and research opportunities. In: International conference on computational logistics. Springer International Publishing, Cham, Switzerland, pp 386–397 Velasquez M, Hester P (2013) An analysis of multi-criteria decision making methods. Int J Oper Res 10(2):56–66 Wagener N (2014) Intermodal transport in Europe—opportunities through innovation. LogForum 10(4):371–382

Chapter 22

Maintenance Service Logistics Joachim Arts , Rob Basten and Geert-Jan van Houtum

Abstract Capital goods, such as manufacturing equipment, trains, and industrial printers, are used in the primary processes of their users. Their availability is of key importance. To achieve high availability, maintenance is required throughout their long life cycles. Many different resources such as spare parts, service engineers and tools, are necessary to perform maintenance. In some cases, e.g. for trains, also maintenance facilities are required. Maintenance service logistics encompasses all processes that ensure that the resources required for maintenance are at the right place at the right time. In a broader sense, it also includes maintenance planning and design-for-maintenance. We first discuss capital goods and the requirements that their users have, which leads us to basic maintenance principles and the structure of typical service supply chains. Next, various relevant decisions and supporting theories and models are discussed. Finally, we discuss the latest developments within maintenance service logistics.

22.1 Maintenance of Capital Goods In this chapter we discuss (maintenance) service logistics for capital goods. Examples of capital goods are manufacturing equipment, trains, industrial printers, radar systems, MRI-scanners, and baggage handling systems. Their availability is of key importance for the users of capital goods, since they are used in the primary processes of the users. Downtime of a capital good may lead to delayed or even lost production, or it leads to inferior service levels and reduced satisfaction at customers. For example, downtime of the bottleneck machines in a semiconductor factory leads J. Arts University of Luxembourg, Luxembourg City, Luxembourg e-mail: [email protected] R. Basten · G.-J. van Houtum (B) Eindhoven University of Technology, Eindhoven, The Netherlands e-mail: [email protected] R. Basten e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_22

493

494

J. Arts et al.

Fig. 22.1 Life cycle of a capital good (from International Organization for Standardization 2008)

to reduced output when the factory produces at maximal capacity (24/7) and then the downtime costs are in the order of tens of thousands of euros per hour (ASML 2013). There also exist situations where downtime or malfunction can lead to safety issues, for example in the case of radar systems on naval vessels or medical equipment in operating rooms. Capital goods may have long life cycles of one to several decades. The life cycle typically follows the seven phases of the model of the International Organization for Standardization (2008), which is presented in Fig. 22.1. In the first three phases, the needs of the users are identified, several ideas for solutions are explored and developed, and finally one is selected that is worked out to a complete design. Next, one or multiple units of the capital good are manufactured in the production phase. These units constitute the installed base. The use phase is the key phase, since the systems are then used for producing products or delivering services. The support phase, which encompasses the maintenance activities, is shown in parallel to this phase. Finally, the systems are removed from operation in the retirement phase. The two key performance indicators of capital goods are its availability or overall equipment effectiveness on the one hand, and its life cycle costs (LCC) or total costs of ownership (TCO) on the other hand. The operational availability is the total time that a capital good is available to produce when required, divided by the total time that the capital good is required to produce. In case of 24/7 usage, this simplifies to the long run average number of available hours per day, divided by 24. When overall equipment effectiveness is used, which is especially popular in the process industry, it also incorporates that a system may not simply be available or not but may be producing at a lower rate: the overall equipment effectiveness is the achieved production capacity divided by the theoretical maximum production capacity. For instance, let us assume that a poultry processing line is used 16 h per day and that it can process 10,000 chickens per hour when it is fully functioning. However, if currently it can only process 120,000 chickens per day, its overall equipment effectiveness is 120,000/(16 × 10,000) = 75%. The LCC are the total of all costs incurred during the seven phases of the life cycle for the entire installed base. This concept is typically used by original equipment manufacturers (OEMs). The TCO are the costs of a particular installed system as observed by the owner, and is closely related to LCC. Typically, TCO consists of acquisition costs at the end of the production phase, use and support costs, and costs (or benefits) related to the retirement of the system. However, more and more, equipment is not purchased but leased, and service contracts are performance based. In such cases, the total cost of ownership is built up differently. We discuss different service contract types in Sect. 22.4.2. Notice that a far-reaching service contract

22 Maintenance Service Logistics

495

may also imply that the OEM (or system integrator) becomes fully responsible for achieving a certain availability of the capital good. Both the LCC and the availability of a capital good are to a large extent determined by its design. As a result, we argue that service logistics in the small sense encompasses all processes that ensure that required resources for maintenance are at the right place at the right time. In the broad sense, it also includes the maintenance concept and design-for-maintenance, i.e., designing capital goods such that their maintenance will be easier and less costly. Let us denote the total downtime of a capital good in a year, say, by D t . It consists of both planned and unplanned downtime, denoted by D p and D u , respectively. Planned downtime results from planned maintenance, which is scheduled P times per year, while unplanned downtime results from failures that happen F times per year on average (F is also called the failure frequency of the system). Unplanned downtime may consist of time to get a service engineer and possibly diagnosis tools at the system or to bring the system to a repair shop, diagnosis time to find the cause of a failure, waiting time for required spare parts and service tools, and the actual time to fix the problem. The diagnosis time and actual time to fix the problem is called the active maintenance time. The other time components constitute waiting time. Also planned downtime consists of actual maintenance time and waiting time. We denote the mean active maintenance time (mean waiting time) for planned and unplanned maintenance by M p and M u (W p and W u ), respectively. We can then state: Dt = Dp + Du,

(22.1)

where D p = P(W p + M p ), D u = F(W u + M u ).

(22.2)

The failure frequency F is a result of the design (and production) and also the active maintenance times M p and M u are largely determined by the design of the capital good: some components are easily exchanged, while others are not. Designfor-maintenance can decrease F, M p , and M u . In order to further keep the active maintenance time low, typically failures are handled by a repair-by-replacement policy in which a failed part is replaced by a spare part. Planned maintenance reduces the failure rate F but increases P. One main advantage of replacing unplanned maintenance by planned maintenance is that it is typically executed much faster. One may be more efficient during the active maintenance time (M p < M u ) and all required maintenance resources can be made available at the time that the planned maintenance starts (W p  W u ). Another main advantage is that users of capital goods can adjust their production plans when maintenance is planned. Hence, having

496

J. Arts et al. Maintenance strategies

Modificative maintenance

Preventive maintenance

Breakdown corrective maintenance

Usage based maintenance

Component replacement and/or overhaul

Condition based maintenance

Block replacement and/or overhaul

Condition monitoring

Periodic inspections

Fig. 22.2 Overview of maintenance strategies (from Arts 2017)

planned downtime is preferred over having unplanned downtime. Service logistics in the small sense can reduce the waiting times W p and W u . Figure 22.2 gives an overview of maintenance strategies. Modificative maintenance concerns interchanging a component with a technologically superior component or a component with a lower failure rate (i.e., with longer times between failures). Often, this is not considered to be a maintenance strategy and we will also ignore it in the sequel. The other maintenance strategies are preventive and breakdown corrective maintenance. Under a breakdown corrective maintenance strategy, a part is replaced only after failure, while under a preventive maintenance strategy, the aim is to replace a part before failure. Of course, a part may still fail before it is replaced. Breakdown corrective maintenance is an attractive option for components that do not degrade or for which the degradation behavior is unclear, such as most electronic components. Such components generally have constant failure rates (also called hazard rates) and the distribution function of the time-to-failure can be modelled with an exponential distribution. This distribution is memoryless, which means that the probability of failure in the next period (a day, say), is independent of its age. This implies that replacing a functioning part by a new part does not bring any benefits. For parts that do wear, it may be beneficial to follow a preventive maintenance strategy. These parts have a so-called increasing failure rate, meaning that over time,

22 Maintenance Service Logistics

497

the probability of failure in the next period increases. The time-to-failure can then be modelled with a Weibull distribution with shape parameter β > 1, and it may then be worthwhile to replace a part by a new part after some time or usage duration. There exist two preventive maintenance strategies: usage based maintenance and condition based maintenance (CBM; this type of maintenance is also called predictive maintenance). Under usage based maintenance, the total usage duration of a part is measured and maintenance is conducted when a certain threshold level has been reached. The usage duration of parts can be measured as time in the field (the corresponding maintenance policy is then also called age based maintenance), number of landings for an airplane, or number of sheets printed for an industrial printer, for example. Since the usage of the capital good is typically planned, also the moment that maintenance is to be performed can be planned. If there is a large set-up cost associated with maintenance, it may be beneficial to interchange several parts simultaneously, which leads to a block replacement and/or overhaul. Otherwise, maintenance can be performed on a single component. Barlow and Hunter (1960) developed the first usage based maintenance models, and new models are still being developed (see, e.g., Arts and Basten 2018). In CBM, the actual condition of a part is monitored, e.g., a vibration level, the number of metal particles in lubrication fluid, or a temperature. Depending on the nature of the capital good or the component, the condition can be measured either continuously through a sensor or periodically during inspections. The simplest form of CBM is to trigger maintenance when the measured condition exceeds a predetermined threshold, e.g., when the temperature rises above 90 ◦ C. If there is a set-up cost associated with maintenance, for example because a service engineer needs to go to a customer or because the capital good needs to be shut down for maintenance, it may be beneficial to interchange several parts simultaneously (as in the block replacement policy discussed above).

22.2 Basics of Maintenance Service Logistics (Basic) 22.2.1 Service Networks Many different resources are required to perform maintenance, such as spare parts, service engineers, and tools. In some cases, e.g. for trains, also maintenance facilities are needed. To ensure that the required resources are at the right place at the right time, a service network or service supply chain is used, which consists of stock locations for spare parts, repair shops, possibly maintenance facilities, and locations of service engineers. The involved costs are huge. All commercial airlines together are estimated to have over $40 billion worth of spare parts (Harrington 2007). A single

J. Arts et al.

# Capital goods / # Spare parts

498

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Year in the life cycle Installs

Retirements

Installed base

Spare part sales

Fig. 22.3 Life cycle of the installed base

company such as ASML, which builds lithography equipment used in semiconductor manufacturing, owns spare parts worth tens of millions of euros (Kranenburg 2006), and the US Coast Guard Aircraft owns inventories worth over $700 million (Deshpande et al. 2006). The demand intensity for spare parts changes during the life cycle of the installed base. Figure 22.3 shows this life cycle (not to be confused with the life cycle of one capital good, as shown in Fig. 22.1). We discuss the design of service networks in Sect. 22.3.1. We typically observe two types of service networks in practice, user networks and OEM networks. User networks are still encountered often, although maintenance activities are increasingly outsourced to OEMs. A good example is the military industry, where generally most maintenance is performed by military organizations themselves, despite the increased complexity of the equipment. This is partly due to the belief that a military organization should not be dependent on civilian support. Therefore, a typical example of a user network is the spare parts network of a military organization; the spare parts networks in the high-tech industry (e.g., computers, printing equipment, medical systems) provide typical examples of OEM networks. Both types of networks often consist of two echelon levels, i.e., there are stocking locations at the local level and at a central level, but networks may consist of more echelon levels. We list the characteristics of both types of networks in Table 22.1 and show an archetypical example of a user network and an OEM network in Figs. 22.4 and 22.5 respectively. While these networks mainly represent the stocking locations for spare parts, the network is quite similar for expensive tools. For the location of service engineers, multiple variants may occur. An OEM that sells technically advanced systems and takes care of most of the maintenance, typically has first line service engineers located relatively close to the installed systems, while second line service engineers are located at a central level. For less advanced systems, a customer may do a large part of the maintenance. Then the OEM may still be asked to deliver spare parts and possibly service tools,

22 Maintenance Service Logistics

499

Table 22.1 Characteristics of two network archetypes (cf. Basten and Van Houtum 2014) User networks (typical for military systems) OEM networks (typical for high-tech systems) Preventive maintenance dominates Two echelon levels in one region No emergency option Mainly internal repair shops Relatively loose service targets

Corrective maintenance dominates Global network with two echelon levels Emergency option at highest echelon level External and internal repair shops Strict/high service targets

Fig. 22.4 Archetypical network of a user that maintains its own systems (cf. Basten and Van Houtum 2014)

but service engineers of the OEM may only be needed in exceptional cases. In that case, service engineers will only be located at a central level. In the remainder of this chapter, we use the terminology that is common in an OEM network setting, and we focus our discussion on that setting. However, most of what we discuss is also applicable for user networks.

Fig. 22.5 Archetypical network of an OEM that maintains its sold systems (cf. Basten and Van Houtum 2014)

500 J. Arts et al.

22 Maintenance Service Logistics

501

22.2.2 Case Study: ASML ASML is a Dutch equipment manufacturer in the semiconductor industry. ASML designs, supplies and integrates advanced technological equipment, called lithography machinery. The customers of ASML (such as Intel, AMD or Samsung) have billion dollar plants that house and exploit the equipment to produce semiconductors. The lithography step is one of many steps in the hightech production environment, but this step is crucial in terms of throughput and in terms of ‘chip generation’. In semiconductor plants, the ASML machines are the most capital intensive, which generally means that plants are developed such that ASML machines are the bottleneck of the process. Customers therefore generally strive for high utilization rates of the machines in their production processes. If an ASML machine breaks down, this can cause the whole production process at the customer to go down (cf. ESCF 2012). As already stated above, downtime can result in costs in the order of tens of thousands of euros per hour. ASML has service contracts with its customers in which an availability level of the machines is agreed upon. It is the responsibility of ASML to ensure that the machines sold are on average up and running for, for instance, 95% of the time. The downtime of the machines can be divided into downtime due to preventive maintenance (planned downtime) and downtime due to corrective maintenance (unplanned downtime). How the total yearly downtime is divided into planned and unplanned downtime can be decided by ASML. Internally at ASML, it has been decided to divide the target for the total downtime into separate targets for planned downtime and unplanned downtime. The unplanned downtime is further divided into targets for the active maintenance time and waiting time. Meeting the targets for unplanned downtime is most challenging. ASML has positioned spare parts, service engineers, and service tools in a global service network consisting of central and local customer service points. ASML has dozens of local support offices, including warehouses, around the world near its customers, and a few central warehouses (see Vliegen 2009).

22.2.3 Decisions and Decision Levels An OEM which provides maintenance services for systems installed at many different customers is faced with many different decisions at a strategic, tactical, and operational level. We describe the main decisions below, and an overview is displayed

502

J. Arts et al.

Table 22.2 Main decisions at the strategic, tactical, and operational decision level Strategic Tactical Operational Product design Maintenance concept

Inventory levels spare parts Inventory levels service tools

Portfolio maintenance services Number of service engineers Design service process Design service network

Lead times repair loops Batching decisions

Control concept repair loops

Parameters of maintenance policies

Allocation rules spare parts Allocation rules service engineers Scheduling of service engineers Priorities repair loops Proactive movements of parts and tools Actions for specific customers

in Table 22.2. We discuss approaches and/or decision support models for different decisions and refer to subsequent sections for more information where appropriate. At the strategic level, the requirements regarding system availability should be taken into account in the product design (see also Sect. 22.4.2). For example, in order to meet very high system availability requirements, one can decide to install more reliable components, to build in redundant components, or to install extra sensors to enable predictive maintenance for certain components. The maintenance concept has to be decided immediately after, or, preferably, in parallel with the product design. This is also related to the portfolio of maintenance services that an OEM wants to offer, including the pricing of those services (Sect. 22.2.4). One needs clear standard service elements (regarding spare parts, service engineers, training of personnel of customers, technical documentation, and so on) with corresponding service processes of the OEM, and customers may be offered service contracts with varying combinations of service elements. An example of a standard service element is that the OEM guarantees spare parts deliveries within 1 week. Another service element may be that spare parts can be delivered within 24 hours in urgent situations, but one has to decide which customers can use that element. The OEM can decide that this is only possible for customers that have this element included in their service contract. Alternatively, the OEM may charge a higher price for this element for customers without a service contract. Another strategic decision concerns the locations of spare parts stocks and stocks of service tools and the division of the world in service areas with pools of service engineers (Sect. 22.3.1). The OEM also has to decide what procedures are in place when a part, tool or engineer is not available at the requested location. Procedures for emergency shipments of parts and tools are common in many service networks, but they need to be established before they can be used (to prevent ad hoc decisions as much as possible). At the tactical level, the OEM has to decide on how many spare parts and service tools have to be kept on stock at each of the locations in the service network. When demand levels are low, it is common to follow base stock policies, under which inventory positions are kept at constant levels. When demand levels are somewhat

22 Maintenance Service Logistics

503

higher, it may be appropriate to include batching rules for replenishments of (new) parts at the central warehouse and possibly within the service network and for placing repair orders for failed parts. The OEM also has to decide on the number of service engineers per service region and the planned lead times for repairable spare parts that are repaired either at repair shops of the OEM itself or at external suppliers. While the maintenance concept has been decided at the strategic level, including the type of maintenance that is followed for the various components, the OEM or user can decide on the parameters of the maintenance policies at the tactical decision level. At the operational level, an OEM first of all plans to follow the rules that have been determined at the tactical level. But, one may be faced with various types of deviations of what has been assumed there such as delayed deliveries of repaired spare parts by repair shops, exceptionally high demand rates for some parts, and long-term illness of multiple service engineers in the same service region. Further, at an operational level more detailed information is available which the OEM wants to take into account. Hence, allocation decisions have to be taken for spare parts and service tools, and, based on actual inventory levels, it has to be determined which parts should get priority in the repair loops. Further, scheduling of service engineers is arranged at that level. When condition information of critical components is monitored and one receives an indication of an upcoming failure of a specific component at a specific location in the world, a required spare part and possibly service tools may be moved closer to that specific location. The OEM also has to monitor continuously whether agreements in the contracts with its customers are being met. If, e.g. due to a temporary lack of resources in the first part of a contract period for a specific user, the OEM is not on schedule to fulfill the agreements for the whole contract period, additional resources may be allocated to a location close to that user’s systems in the remaining contract period.

22.2.4 Service Portfolio and Service Processes Many different maintenance services are possible for capital goods. In service contracts, agreements may be specified regarding preventive, breakdown and modificative maintenance, and they can cover spare parts, service tools and service engineers. Contracts can also be closed for any subset of these elements. For each element, it has to be specified how quickly that element is provided and its price has to be determined. Different customers may ask for different speeds with which a service is delivered. An OEM has to carefully determine its service portfolio, in order to prevent too many different types of agreements with customers, which will lead to an unmanageable situation. For breakdown maintenance services for large-scale computers, service contracts are offered with response time targets of 2, 4, 8 h, next business day, and second next business day. The response time starts when the customer calls the OEM to report a failure of the system. The call is received at a call center from where a service operator will try to solve the problem remotely. If this is possible, the problem will be solved relatively quickly. If a failed part has to be replaced by

504

J. Arts et al.

a spare part, then that part will be delivered by a fast delivery from one of the local warehouses in the service network. The part has to arrive at the customer within the agreed response time. Some parts are replaced by the customer itself, while other parts have to be replaced by a service engineer. The customer agrees with the local service organisation when the service engineer will replace the required spare part. The customer has the right to insist on the availability of a service engineer within the target response time, but he may also schedule the visit of the service engineer some time later. This depends on which part failed (for some parts, there is redundancy in the system, e.g. for hard disks) and on which processes are currently running. In the above example, it becomes clear that response time is an appropriate service measure to base agreements on. Generally, which service measure is appropriate depends on the interaction between the customer and the OEM. A customer desires a smooth production or service process, and the OEM wants an efficient maintenance process. An important question is how strong the functioning of the system is related to the production or service process of the user. If failures of systems are strongly dependent on what production or service processes are executed on the system, on the utilization rate, or on the behavior of the operators, then this has to be taken into account when defining agreements in service contracts. One may then decide to base agreements on response times (as in the above example) or on the total solution time per failure (consisting of waiting and repair time) rather than on system availability itself which also depends on the number of failures per time period (see (22.1)). We see different types of service measures in different industries. The use of response times and solution times is common for large-scale computers, printing systems and medical equipment. In the semiconductor industry, it is common to use the yearly or monthly downtime due to waiting for spare parts, which includes the effect of the number of failures.

22.3 Modeling and Decision Making (Advanced) 22.3.1 Design of a Service Network Consider an OEM with a service network of the type depicted in Fig. 22.5. Suppose that the systems have to be maintained at the places where they are installed; this is common for e.g. manufacturing equipment. The OEM typically divides the world in regions that may coincide with one or more continents (e.g., Europe, Asia-Pacific, Americas). Per region, there is one central warehouse and multiple local warehouses for the spare parts and service tools (cf. Fig. 22.5). Further, there are subregions (e.g., countries or parts of countries) with a pool of service engineers per subregion. These service engineers execute all first line maintenance services for the systems installed in their subregion. Service engineers may travel by car and may have a stock of spare parts in their car. In that case, these car stocks can be seen as local stockpoints as well. In several networks in practice, engineers have no car stocks, in which case a

22 Maintenance Service Logistics

505

weaker coupling is obtained between the planning of spare parts and service tools on the one hand and the planning of service engineers on the other hand. For situations without car stocks and with tight targets for response or solution times, as in the large-scale computers example in Sect. 22.2.4, the choice of the locations for the central and local warehouses may be approached as a facility location problem. In this problem, customers are assumed to be distributed over a large area and they generate demands for spare parts and possibly service tools. There may be many candidate locations for the warehouses. Each customer has to be connected to a local warehouse that is at a sufficiently close distance to deliver parts/tools within x h, where x is such that the highest level of service can be realized (we assume that this level has to be realized for many customers, but the set of customers with the highest service level may vary over time). Costs that are incorporated are fixed costs per local warehouse, the costs for the flows of the spare parts and the service tools, and inventory holding costs for spare parts and service tools. This leads to various types of mixed integer linear programming problems; see e.g. Candas and Kutanoglu (2007) and Rappold and van Roo (2009). Other related decisions that have to be taken are: (i) which components should be repaired and which should be discarded (repair or buy); (ii) where in the service network should parts be repaired; (iii) where should the manpower and equipment needed for these repairs be installed? These decisions can be addressed by so-called Level Of Repair Analysis (LORA) models; see e.g. Basten et al. (2015).

22.3.2 Forecasting Planning service logistics inevitably starts with forecasting the demand for all resources needed for maintenance activities. Resources such as engineers and certain tools and parts are used often and then forecasting is not very different from traditional forecasting problems. However, many spare parts and also advanced service tools are used only sporadically. Then forecasting based on historic time series data (e.g., monthly demands) is difficult, because such data consists of many zeros. Such demand is also called intermittent demand. The usual forecasting techniques based on time series analysis such as exponential smoothing are ill-suited to deal with intermittent demand. To deal with intermittent demand, time series forecasting methods have been developed that explicitly forecast the expected number of periods with zero demand. This forecast is then combined with a forecast for the demand quantity based on only periods with non-zero demand. These ideas have been pioneered by Croston (1972) and notable extensions have been made by Syntetos and Boylan (2005), who show how to reduce bias in this forecasting technique, and Teunter et al. (2011), who show how to update the forecast for the time between positive demand in periods with zero demand. Machine learning algorithms such as neural networks have also been designed to forecast intermittent spare parts demand based on time series data as well as the more traditional bootstrapping and Box-Jenkins methods.

506

J. Arts et al.

Another way to forecast intermittent demand for spare parts and service tools is to link the demand forecast to covariates such as planned maintenance and inspections. (Forecasting based on covariates is also called causal forecasting.) Rather than forecasting the demand directly, maintenance and inspections are forecasted. These forecasts are then linked with a forecast of the probability that such an inspection will lead to demand. These methods require knowledge on maintenance programs and planning, but yield more accurate forecasts. Wang and Syntetos (2011) and Romeijnders et al. (2012) show how this idea works in different settings. For a situation with components that rarely fail in a small installed base, it is also useful to estimate the remaining useful lifetime of a component and construct a forecast based on that. Such remaining useful lifetime models are often technology specific, see e.g. Jardine et al. (2006) for the prevalent technology of rotating machinery. Bian et al. (2015) provide a Bayesian framework to update residual lifetime distributions based on measurement from sensors.

22.3.3 Inventory Control of Spare Parts As denoted in Fig. 22.3, the demand pattern of spare parts consists of three phases: a first, initial phase, a second phase with stationary demand, and a third, end-of-life phase. In all phases, the aim is to choose the inventory levels of spare parts such that agreements in service contracts can be met. These agreements are in terms of system availability, downtime or closely related measures. This leads to the formulation of multi-item inventory models in which relevant costs are minimized subject to constraints on or related to system availability. The relevant costs typically consist of inventory holding costs and transportation costs for replenishments, emergency shipments, and lateral transshipments. The system availability constraints may be formulated for local warehouses and/or groups of installed machines. Solving these multi-item models requires a system approach, as introduced by Sherbrooke (1968). Let us illustrate the main idea of the system approach by the following example. Consider a local stockpoint at a production site where multiple machines are installed. The machines have two critical components that are subject to failures. These components are called Component 1 and Component 2. For all machines together, Component 1 fails 6 times per year on average and Component 2 fails 0.6 times per year on average. These failures occur at random times (i.e., we assume that the underlying demand processes are Poisson). When a failure of a machine occurs, the component that has failed can be identified quickly. Next, the failed component is removed from the machine and replaced by a corresponding spare part. If this spare part is in stock, the replacement can be carried out immediately, within a very short time, and the downtime is negligible. In that case, the failed part is sent to a repair shop and, once repaired, will be returned to the local stock point as a ready-for-use part which takes a lead time of 2 months (on average). If there is no spare part on stock, then the failed part is sent to the repair shop for an emergency repair, and the repaired part will be returned as a ready-for-use part after 3 days (on average). In

22 Maintenance Service Logistics

507

Machines Component 1 Cheap

Repair loop

Component 2 Expensive

1

2

1

2

1

2

1

2

Fig. 22.6 Stocking of spare parts at a single production site

that case, the waiting time is 3 days, and we also have to pay an extra-high price for the repair at the repair shop. We assume that the machines are used for many years. Hence, we assume an infinite horizon and consider the performance in the long run, i.e., in steady state. Figure 22.6 shows a visualization of this example. Let us assume that S1 spare parts of Component 1 and S2 spare parts of Component 2 are initially put in stock. Then, under the assumptions made above, we have a closed loop for the spare parts, and the stock on hand of Component i plus the number of parts in repair is always equal to Si , i = 1, 2. The initial stock levels are our decision parameters, and we seek to minimize the initial stock investment subject to an unplanned downtime constraint. The price per new spare part of Component i is denoted by ci , i = 1, 2, where c1 = 100 euros and c2 = 10, 000 euros. Let βi (Si ) be the probability that there is a ready-for-use part on stock when an arbitrary failure of a Component i occurs. The mean unplanned downtime per year (for all machines together) is then equal to (cf. (22.2)) D u (S1 , S2 ) = 6 · (1 − β1 (S1 )) · 3 + 0.6 · (1 − β2 (S2 )) · 3

(in days).

Let us assume that this mean unplanned downtime per year may be at most 2.0 days. Then, our decision problem is as follows: (P)

min

c1 S1 + c2 S2

subject to D u (S1 , S2 ) ≤ 2.0. Notice that the downtime constraint ensures that the yearly number of fast repairs (= 6 · (1 − β1 (S1 )) + 0.6 · (1 − β2 (S2 ))) is limited and hence the extra costs for these repairs do not have to be included in the objective function. Further, the yearly mean number of repairs does not depend on the initial spare parts stocks and thus

508

J. Arts et al.

their costs are constant. As a result, the objective function only consists of the initial stock investment. The performance of a given solution (S1 , S2 ) is obtained by a steady-state analysis of the inventory system and the optimal solution can be obtained by enumeration for this small problem (see also Van Houtum and Kranenburg 2015, Chap. 2). The optimal solution appears to be (S1∗ , S2∗ ) = (5, 0) and the corresponding performance is: Initial stock investment = 100 · 5 + 10, 000 · 0 = 500 euros, D u (S1∗ , S2∗ ) = 6 · (1 − β1 (S1∗ )) · 3 + 0.6 · (1 − β2 (S2∗ )) · 3 = 1.86 days. So, this is the performance that we obtain by following a system approach. Now suppose that, for some reason, we do not want to follow the system approach. In that case we model the unplanned downtime constraint at component level such that both initial stock levels can be optimized individually. Then the ‘budget’ of 2.0 days for the unplanned downtime first has to be divided over the two components. We can do this proportionally to their demand rates, which gives a budget of (6/6.6) · 2.0 = 1.82 days for Component 1 and a budget of (0.6/6.6) · 2.0 = 0.18 days for Component 2. The problem to be solved for Component 1 is as follows: (P1 )

min c1 S1 subject to 6 · (1 − β1 (S1 )) · 3 ≤ 1.82.

and a similar problem is obtained for Component 2. We then obtain S1∗ = 3 as optimal solution for Component 1 and S2∗ = 1 for Component 2. The resulting performance at machine level is then equal to: Initial stock investment = 100 · 3 + 10, 000 · 1 = 10, 300 euros, D u (S1∗ , S2∗ ) = 6 · (1 − β1 (S1∗ )) · 3 + 0.6 · (1 − β2 (S2∗ )) · 3 = 1.62 days. This nicely meets the unplanned downtime constraint, but the initial stock investment is 20 times as high (10,300 vs. 500 euros)! This is purely due to the unnecessary decomposition of problem (P) into single-component problems. The system approach avoids such a decomposition and searches for a solution in a much larger solution space. The theory on multi-item inventory models and the system approach are well developed. This theory is focussed on the second, stationary phase for spare parts, when generally reliable forecasts can be made for the demand rates and a steadystate analysis is appropriate. For inventory control in the first, initial phase and third, end-of-life phase, much less theory is available. In the initial phase, it may be difficult or even impossible to create reliable forecasts. Companies often use pragmatic approaches in that phase. Hardly any literature is available for this phase; for two recent contributions, see Martinetti et al. (2017) and Van Wingerden et al. (2017). In the third, end-of-life phase, also a separate approach is required. For many

22 Maintenance Service Logistics

509

parts, it is not possible anymore to produce or buy new parts after some time point. At that moment, a last buy has to be made or one has to develop another sourcing possibility; see Pourakbar et al. (2012) and Behfard et al. (2015), and the references therein.

22.3.4 Repair Loops Repair-by-replacement significantly decreases the downtime of capital goods after a failure. Many replaced components are sufficiently expensive (and large) to warrant off-line repair. Such components are therefore called repairables. After being repaired, repairables are returned to stock as ready-for-use parts. Service logistics networks therefore consist of closed loops of repairable spare parts and repair shops where such parts are repaired or refurbished. Each repair shop in a service supply chain needs to be managed and this requires addressing the following three questions: (i) how many repair resources are needed; (ii) how should repairs of failed components be scheduled; (iii) how many lower level parts need to be stocked to facilitate the repair of components? We will discuss the main considerations and methods involved in making these decisions. The sojourn time of a job in a repair shop consists mostly of waiting time for resources, such as technicians, tools, and lower level parts. The number of repair resources in a repair shop determines repair lead times that inventory control needs to account for. The number of resources such as technicians and tools in a repair shop can be dimensioned by using techniques from queueing theory. The fundamental machine-repair problem in queueing theory is a stylized model that was conceived for exactly this purpose. Taylor and Jackson (1954) designed this model for the engine repair shop of British Airways. Each repair shop will have specific characteristics and many variations to this queueing model have been developed for a variety of settings; see Koenigsberg (1982). Scheduling in repair shops is done with the objective to avoid downtime of capital goods by preventing stockouts of repairable components. This is different from scheduling in production settings where the objective is to meet customer due dates or production efficiency targets. Repair shop scheduling rules should therefore take current stock levels of repairables into account, as well as shop characteristics and information. Hausman and Scudder (1982) test several scheduling rules in a simulation study and find that using smart scheduling rules yields a performance increase which allows a repairable item inventory reduction of 20% relative to a system with first come first service scheduling with equivalent performance. One or more lower level parts are usually needed to complete the repair of a component. Stocking these lower level parts—which are also called Shop Replaceable Units (SRUs)—is therefore an important decision in repair shops. Muckstadt (1973) shows how inventory of SRUs and repairables can be controlled jointly when a component always breaks down due to exactly one SRU. In practice it is important to deal with the case where components need multiple SRUs to complete a repair. The

510

J. Arts et al.

inventory control of SRUs is then quite similar to the inventory control of assembleto-order systems where it is also uncertain which parts will be needed to fulfill an order. Van Jaarsveld et al. (2015) provide methods to minimize holding inventory of SRUs subject to the requirement that all parts needed for repair of an arbitrary component will be available with a high probability within a given time window.

22.3.5 Service Tools When a machine fails, one or multiple service tools may be needed to solve the failure. This includes diagnosis and calibration tools. If the number of different tools is low and their prices are low as well, then each service engineer may be given a kit with all relevant tools. However, service tools can also be expensive and/or there may be many different tools. In that case, the problem for service tools is similar to the problem for spare parts: Where should tools be stocked and how many of them should be stocked? When a machine fails, one aims to send the required service tools together with the spart part(s). The planning of the service tools may be done independently from the planning of the spare parts. However, a machine failure can only be solved when both the required spare part(s) and service tool(s) are available. Hence, planning them in an integrated way is better, in particular when the same combinations of parts and tools are often needed. Let us further elaborate on what is needed for an integrated planning. First of all, it requires that the repair process is well structured. We need to have a list of all possible machine failures and a standard repair procedure per machine failure. The required part(s) and tool(s) per machine failure need to be known as well. Further, which machine failure occurred must be known already at the moment that the part(s) and tool(s) are sent to the failed machine. In addition, we need data on how often each machine failure occurs. If all these requirements are met, we can model coupled demands for parts and tools (as in assemble-to-order systems) in a multiechelon network and this leads to complicated inventory problems. This integrated planning problem has been studied by Vliegen (2009). She studied this problem in a single-echelon, multi-location setting with lateral and emergency shipments (see Vliegen 2009, Chap. 6). She developed a heuristic solution procedure that takes the coupling in demands into account, but the computational complexity of this heuristic is relatively high. It is a great challenge to develop an efficient heuristic that works for instances of real-life size and that can be extended to multi-echelon networks with lateral and emergency shipments.

22 Maintenance Service Logistics

511

22.3.6 Service Engineers The maintenance of capital goods is executed by service engineers. Depending on the capital good, the service engineer travels to the capital good (e.g. manufacturing equipment) or the capital goods travel to the service engineer (e.g. aircrafts). In both cases, it is important to determine how many service engineers are needed so service all capital goods in a given region, and in the former situation, also the dispatching and routing of engineers to sites where capital goods need maintenance is important. The number of service engineers that is available to respond to calls directly affects the waiting times W p and W u for maintenance in Eq. (22.1). Queueing theory can be used to determine the waiting time that results when a given number of service engineers need to handle all calls in a given area. Since this waiting time is also affected by the availability of spare parts, it is useful to consider stocking of spare parts and staffing of service engineers jointly; see Rahimi-Ghahroodi et al. (2017). When engineers travel to capital goods, the travel times are also part of the waiting time. The waiting time due to travelling of the service engineer can be affected by dispatching and routing decisions. However, dispatching and routing decisions also affect the effective available capacity of service engineers, because a service engineer is also “utilized” when travelling and the cumulative travel time that engineers experience is affected by routing and dispatch decisions. This interaction is studied in some more detail by Agnihothri and Karmarkar (1992). Dispatching and routing decisions can also be affected by the availability of additional information through condition monitoring. The available condition information can be used to predict when and where a service engineer will be needed and this information in turn can be used advantageously in routing and dispatching; see Ichoua et al. (2006). Dispatching and routing of emergency services such as ambulances have been leading in developing methods to support decision making in this field; see e.g. Maxwell et al. (2010).

22.4 New Developments in Maintenance Service Logistics (State-of-the-Art) 22.4.1 Technological Developments The developments in sensor technology make it possible to obtain much more information on the degradation of components against a low cost. In addition, the developments in communication technology enable an OEM (or a third party) to collect degradation and other relevant data in one place in the world (or in the cloud). This gives a number of opportunities: • Consider a component for which the prevalent failure mode is known and for which it is known how to measure the condition. By analyzing the data for many (failed)

512

J. Arts et al.

components, the OEM may be able to improve the modeling of the degradation process and hence the measurement of the condition. The OEM may be able to improve the threshold that triggers maintenance. • Consider a component for which the failure mode is not known or the failure mode is known in principle but it is not known how to measure the condition directly. Then data of many systems can be analyzed by regression or data mining techniques in order to construct failure predictors. • For quite some components, it is known that their degradation process also depends on the environment or on the type of products that is being produced. This may be analyzed and can be incorporated in the failure prediction. Generating a failure prediction is only useful when the prediction occurs early enough to take a preventive measure. For quite some components, a failure develops itself very quickly or even happens suddenly. In that case, CBM is not possible and one can stick to breakdown maintenance. For some failure predictions, the OEM can be 100% sure that the failure will happen and there is a good prediction of the time at which the failure will occur. In that case preventive maintenance makes sense. If there is uncertainty about when the failure occurs, too much useful lifetime may be lost if a preventive maintenance action is applied to early. But the OEM can already ensure that the failure can be fixed quickly when the failure happens. E.g., if the failed component has to be replaced by a spare part, then the availability of a spare part can be assured (so that the waiting time is low; see (22.1)). For failure predictions that are generated by regression analysis or data mining, the OEM has to make a tradeoff between false positives (a failure is predicted but will not occur) and false negatives (no prediction of a failure, but a failure does occur). Executing preventive maintenance based on unreliable predictions may be unattractive. Nevertheless, still such predictions can be used to initiate further investigations or to ensure that the failure can be fixed quickly when the failure does happen. This may be done by temporarily moving an extra spare part to a location. Topan et al. (2018) show that this can decrease costs significantly, even under a high percentage of false positives among the predictions. In these situations, high downtime costs due to waiting for spare parts are avoided by temporary placements of a required spare part in a local warehouse where normally that spare part is not kept on stock. Another relevant technological development is additive manufacturing. Currently, many spare parts are kept on stock in a local warehouse while demand rates are very low at those local levels. The spare parts need to be kept on stock locally so that long waiting times are avoided when failures occur. It would be ideal if parts do not have to be kept on stock but if they can be quickly produced by additive manufacturing when a failure occurs. Currently, we see multiple factors that limit the use of additive manufacturing: (i) more complex components cannot be produced by additive manufacturing; (ii) applying additive manufacturing for a component requires a costly design, like for traditional manufacturing (this forms a barrier for using additive manufacturing in addition to traditional manufacturing); (iii) the parts produced by additive manufacturing may have a lower reliability. When additive

22 Maintenance Service Logistics

513

manufacturing develops further, these factors may become less limiting. Current research investigates when additive manufacturing is a good option; see Knofius et al. (2016), Song and Zhang (2017), and Westerweel et al. (2018).

22.4.2 New Business Models for Maintenance Services We already discussed maintenance services that can be offered by OEMs (or third parties) to users of capital goods. We see for many years already that more and more maintenance task are taken over by OEMs. In the ultimate case, the OEM does not sell the system but offers the function of the system and the customers pay for the usage and the system availability. Rolls Royce does this already for a long time for engines of airplanes (Swartz 2014.) Rolls Royce offers so-called power-bythe-hour contracts under which Rolls Royce keeps the ownership of the engines and customers pay based on the number of hours that they use the engines. Other examples include “pay per click” systems offered by professional copier manufacturers and cloud computing (e.g. IBM offers data center capacity). Obviously, also high system availabilities are guaranteed. This movement of offering the function instead of the system is known as servitization. The shift to servitization for advanced capital goods has a number of advantages. The OEM (or a third party) will be responsible for the maintenance of many systems and will be much more efficient than maintenance departments of individual customers with respect to having well-trained service engineers, having specialists for the more complicated problems, and keeping spare parts and service tools on stock at a close distance of installed systems. Also they will be better able to develop advanced maintenance concepts, including CBM and the usage of data mining to generate failure predictions (this is only possible if data of many systems are available). Also, the OEM will be better able to realize a high system availability than an individual user. Further, under a power-by-the-hour contract, a customer knows directly what a system costs per unit of production or service. Another advantage is that an OEM is incentivized to design a more reliable product. When high system availabilities are required, the OEM may decide to use more reliable components or to build in extra sensors to facilitate CBM. An OEM may also think of tailored designs for different market segments with different system availability requirements. And, the OEM can use materials that can be re-used easily when installed systems are finally disposed of. Several of the advantages that are obtained under power-by-the-hour contracts hold also for full service contracts under which the customer still buys the system but the OEM is responsible for almost all maintenance activities and for realizing high system availabilities.

514

J. Arts et al.

22.4.3 Control Towers Suppose that an OEM (or third party) closes power-by-the-hour and full service contracts with many customers and suppose that high system availabilities are required. Then using CBM or condition-based movements of spare parts will help to provide the required services in an efficient way. We note that decisions can be taken at the operational planning level based on the latest information. This may be realized by creating a control tower, like the control towers used in aviation, where the movements of all airplanes are followed and decisions are taken on the routes of airplanes, the order in which they can land at an airport, and so on. In a service logistics control tower, an OEM can bring multiple types of relevant information together: information on all service contracts, the performance for each contract until the current moment, the condition of critical components, the current stocks of spare parts, possible problems with the supply of new spare parts, current positions and routes of service engineer, and so on. All this information needs be present in appropriate information systems (the current ERP systems generally do not suffice). Planners can respond to problems quickly based on the most recent information. An appropriate response can be given by simple decision rules or more sophisticated decision support tools. The development of such control towers constitutes a great challenge, but will be key for success.

22.5 Further Reading For various topics discussed in this chapter, we suggest further readings: Maintenance of capital goods: For more literature on usage based maintenance models, we refer to any textbook on reliability engineering or maintenance optimization, e.g., Jardine and Tsang (2006), Pintelon and Van Puyvelde (2006), Ebeling (2001), and Arts (2017). For reviews on condition based maintenance, see Jardine et al. (2006), Peng et al. (2010), Prajapati et al. (2012), and Olde Keizer et al. (2017). For a review on the topic of clustering of maintenance, see Nicolai and Dekker (2008); their review only includes one paper with CBM. Clustering of components that all use a CBM strategy is a recent topic in the literature, see, e.g., De Jonge et al. (2016) or Olde Keizer et al. (2016). Even more complicated is clustering of components that use different strategies, see e.g. Zhu (2015) and Arts and Basten (2018). Service portfolio and service processes: For different types of service contracts that are used in the industry, we refer to Cohen et al. (2006) and Oliva and Kallenberg (2003). Forecasting: Forecasting methods for intermittent demand based on neural networks and several other machine learning and Box-Jenkins methods are studied by Lolli et al. (2017), Moon (2013) and Pai and Lin (2006).

22 Maintenance Service Logistics

515

Inventory control of spare parts: For textbooks on multi-item inventory models and the system approach, see Sherbrooke (2004), Muckstadt (2005), and Van Houtum and Kranenburg (2015). These books cover multi-echelon networks with a variety of properties that occur in practice, such as the use of emergency shipments, lateral transshipments, multiple customer classes, multiple machine types with commonality, and multiple indenture structures. Various models have been implemented in commercial software tools that are widely used in the industry, but there are also quite some companies with a tailor-made system-oriented solution (for more details, see Basten and Van Houtum 2014, Sect. 8). New business models for maintenance services: For more information on servitization, see e.g. Avci et al. (2015) and Agrawal and Bellos (2017), and the references therein. Acknowledgements The authors gratefully acknowledge the Netherlands Organisation for Scientific Research (NWO) and the Dutch Institute for Advanced Logistics (TKI Dinalog) for their support via the ProSeLoNext project and the NWO-TOP project on “Service Logistics for Advanced Capital Goods”. Joachim Arts also gratefully acknowledges NWO for its support through his Veni grant.

References Agnihothri S, Karmarkar U (1992) Performance evaluation of service territories. Oper Res 40(2):355–366 Agrawal VV, Bellos I (2017) The potential of servicizing as a green business model. Manag Sci 63(5):1545–1562 Arts J (2017) Maintenance modeling and optimization. BETA Working Paper 526 Arts JJ, Basten RJI (2018) Design of multi-component periodic maintenance programs with singlecomponent models. IISE Trans 50(7):606–615 ASML (2013) ASML infographic about supply chain management. Video. https://www.youtube. com/watch?v=dlKqO_L1rdc Avci B, Girotra K, Netessine S (2015) Electric vehicles with a battery switching station: Adoption and environmental impact. Manag Sci 61(4):772–794 Barlow R, Hunter L (1960) Optimum preventive maintenance policies. Oper Res 8(1):90–100 Basten RJI, Van Houtum GJ (2014) System-oriented inventory models for spare parts. Surv Oper Res Manag Sci 19(1):34–55 Basten RJI, Van der Heijden MC, Schutten JMJ, Kutanoglu E (2015) An approximate approach for the joint problem of level of repair analysis and spare parts stocking. Ann Oper Res 224(1):121– 145 Behfard S, Van der Heijden MC, Al Hanbali A, Zijm WHM (2015) Last time buy and repair decisions for spare parts. Eur J Oper Res 244(2):498–510 Bian L, Gebraeel N, Kharoufeh JaiIT (2015) Degradation modeling for real-time estimation of residual lifetimes in dynamic environments. IIE Trans 47(5):471–486 Candas MF, Kutanoglu E (2007) Benefits of considering inventory in service parts logistics network design problems with time-based service constraints. IIE Trans 39(2):159–176 Cohen MA, Agrawal N, Agrawal V (2006) Winning in the aftermarket. Harv Bus Rev 84(5):129–138 Croston J (1972) Forecasting and stock control for intermittent demands. Operation 23(3):289–303 Deshpande V, Iyer AV, Cho R (2006) Efficient supply chain management at the U.S. Coast Guard using part-age dependent supply replenishment policies. Oper Res 54(6):1028–1040

516

J. Arts et al.

Ebeling C (2001) Introduction to reliability and maintainablity engineering, 2nd edn. McGraw-Hill ESCF (2012) Spare parts planning at ASMl. ESCF operations practices. http://www.escf.nl/ operation Harrington L (2007) From just in case to just in time. Air transport world. pp 77–80 Hausman W, Scudder G (1982) Priority scheduling rules for repairable inventory systems. Manag Sci 28(11):1215–1232 Ichoua S, Gendreau M, Potvin JY (2006) Exploiting knowledge about future demands for real-time vehicle dispatching. Transp Sci 40(2):211–225 International Organization for Standardization (2008) Systems and software engineering–system life cycle processes. ISO/IEC/IEEE 15288:2015 Jardine AKS, Tsang AHC (2006) Maintenance, replacement, and reliability. Theory and applications, Dekker Mechanical Engineering, CRC Press, Boca Raton, FL, Jardine AKS, Lin D, Banjevic D (2006) A review on machinery diagnostics and prognostics implementing condition-based maintenance. Mech Syst Signal Process 20:1483–1510 De Jonge B, Klingenberg W, Teunter RH, Tinga T (2016) Reducing costs by clustering maintenance activities for multiple critical units. Reliab Eng Syst Saf 145:93–103 Knofius N, Van der Heijden M, Zijm W (2016) Selecting parts for additive manufacturing in service logistic. J Manuf Technol Manag 27:915–931 Koenigsberg E (1982) Twenty five years of cyclic queues and closed queue networks: A review. J Oper Res Soc 33(7):605–619 Kranenburg B (2006) Spare parts inventory control under system availability constraints. PhD thesis, Eindhoven University of Technology Lolli F, Gamberini R, Regattieri A, Balugani E, Gatos T, Gucci S (2017) Single-hidden layer neural networks forecasting intermittent demand. Int J Prod Econ 183:116–128 Martinetti A, Braaksma AJJ, Ziggers J, Van Dongen LAM (2017) On the initial spare parts assortment for capital assets: a structured approach aiding initial spare parts assortment decisionmaking. In: Redding L, Roy R, Shaw A (eds) Advances in through-life engineering services. Springer, New York, pp 429–442 Maxwell MS, Restrepo M, Henderson SG, Topaloglu H (2010) Approximate dynamic programming for ambulance redeployment. INFORMS J Comput 22(2):266–281 Moon S (2013) Predicting the performance of forecasting strategies for naval spare parts demand–a machine learning approach. Manag Sci Financ Eng 19(1):1–10 Muckstadt JA (1973) A model for a multi-item, multi-echelon, multi-indenture inventory system. Manag Sci 20(4):472–481 Muckstadt JA (2005) Analysis and algorithms for service parts supply chains. Springer, New York Nicolai RP, Dekker R (2008) Optimal maintenance of multi-component systems: A review. In: Murthy DNP (ed) Kobbacy KAH. Complex system maintenance handbook, Springer Series in Reliability Engineering, Springer, London (UK), pp 263–286 Olde Keizer MCA, Teunter RH, Veldman J (2016) Clustering condition-based maintenance for systems with redundancy and economic dependencies. Eur J Oper Res 251:531–540 Olde Keizer MCA, Flapper SDP, Teunter RH (2017) Condition-based maintenance policies for systems with multiple dependent components: a review. Eur J Oper Res 261(1):405–420 Oliva R, Kallenberg R (2003) Managing the transitions from products to services. Int J Serv Ind Manag 14:160–172 Pai P, Lin K (2006) Application of hybrid learning neural fuzzy systems in reliability prediction. Qual Reliab Eng Int 22(2):199–211 Peng Y, Dong M, Zuo MJ (2010) Current status of machine prognostics in condition-based maintenance: a review. Int J Adv Manuf Technol 50:297–313 Pintelon L, Van Puyvelde F (2006) Maintenance decision making. Acco, Leuven (Belgium) Pourakbar M, Frenk JGB, Dekker R (2012) End-of-life inventory decisions for consumer electronics service parts. Prod Oper Manag 21(5):889–906 Prajapati A, Bechtel J, Ganesan S (2012) Condition based maintenance: a survey. J Qual Maint Eng 18(4):384–400

22 Maintenance Service Logistics

517

Rahimi-Ghahroodi S, Al Hanbali A, Zijm WHM, Van Ommeren JKW, Sleptchenko A (2017) Integrated planning of spare parts and service engineers with partial backlogging. OR Spectr 39(3):711–748 Rappold JA, Van Roo BD (2009) Designing multi-echelon service parts networks with finite repair capacity. Eur J Oper Res 199(3):781–792 Romeijnders W, Teunter R, Van Jaarsveld W (2012) A two-step method for forecasting spare parts demand using information on component repairs. Eur J Oper Res 220(2):386–393 Sherbrooke C (2004) Optimal inventory modeling of systems: multi-echelon techniques. Kluwer Academic, Boston/Dordtrecht/London Sherbrooke CC (1968) Metric: a multi-echelon technique for recoverable item control. Oper Res 16(1):122–141 Song JS, Zhang Y (2017) Stock or print? impact of 3d printing on spare parts logistics. Working Paper Swartz E (2014) The transformation of ge: From "we bring good things to life" to industrial machines in the cloud. Blog. http://www.metratech.com/blog/industrial-machines-in-the-cloud/. Accessed 7 May 2017 Syntetos AA, Boylan JE (2005) The accuracy of intermittent demand estimates. Int J Forecast 21(2):303–314 Taylor J, Jackson RRP (1954) An application ofthe the birth death process to the provision of spare machines. Oper Res Q 5(4):95–108 Teunter R, Syntetos A, Babai M (2011) Intermittent demand: linking forecasting to inventory obsolescence. Eur J Oper Res 214(3):606–615 Topan E, Tan T, Van Houtum G, Dekker R (2018) Using imperfect advance demand information in lost-sales inventory systems. IISE Trans Van Houtum GJ, Kranenburg B (2015) Spare parts inventory control under system availability constraints. International Series in OR & MS, Springer, New York Van Jaarsveld W, Dollevoet T, Dekker R (2015) Improving spare parts inventory control at a repair shop. Omega 57:217–229 Van Wingerden E, Tan T, Van Houtum G (2017) Spare parts inventory control under uncertain demand rates. Working Paper Vliegen I (2009) Integrated planning for service tools and spare for capital goods. PhD thesis, Eindhoven University of Technology Wang W, Syntetos A (2011) Spare parts demand: linking forecasting to equipment maintenance. Transp Res Part E: Logist Transp Rev 47(6):1194–1209 Westerweel B, Basten RJI, Van Houtum GJ (2018) Traditional or additive manufacturing? assessing component design options through lifecycle cost analysis. Eur J Oper Res 270(2):570–585 Zhu Q (2015) Maintenance optimization for multi-component systems under condition monitoring. PhD thesis, Eindhoven University of Technology

Part VI

New Developments and Special Topics

Chapter 23

Additive Manufacturing and Its Impact on the Supply Chain Henk Zijm, Nils Knofius and Matthieu van der Heijden

Abstract Additive Manufacturing (AM) is rapidly gaining interest as a highly innovative manufacturing technology, having many advantages over more conventional manufacturing methods. These advantages include the ability to produce very complex structures that are relatively easily customized to specific user requirements. The fact that AM services become affordable for small companies or even for consumers offers possibilities for decentralized manufacturing, downstream in the supply chain. In addition, AM allows for high degrees of flexibility, both in product design and manufacturing, as a result of using smart CAD systems that may be based on accurate scanning technologies. The ability to work with low setup times and costs and to largely eliminate work in progress inventories while maintaining a high degree of supply chain responsiveness makes AM a promising alternative for low-volume, high-value items. In this chapter, we outline the basics of AM technologies, after which we discuss at a more advanced level its impact on the supply chain. Next, we turn to spare parts delivery in after-sales service supply chains; these slow moving parts are often mentioned as ideal candidates for AM. In a state-of -the-art report, we provide a methodology for the identification of spare parts that may appear promising candidates for the application of AM. We conclude with a field study conducted at a service provider in the aerospace industry.

H. Zijm (B) · N. Knofius · M. van der Heijden Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, Overijssel, The Netherlands e-mail: [email protected] N. Knofius e-mail: [email protected] M. van der Heijden e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_23

521

522

H. Zijm et al.

23.1 Additive Manufacturing (Basics) Additive Manufacturing (AM) is a technology that enables the production of complex geometries and near-net shape components. Its name stems from the fact that it builds a component, part or product from raw materials layer by layer (additively). Conventional machining methods such as cutting, milling, drilling or lathing, remove material and are therefore classified as subtractive manufacturing. Although only recently known to the broader public, the first AM technology was already commercialized in the late 1980s, when it was used as a technique for rapid prototyping, called stereolithography. In this technology, a vat with a vertically moving platform is filled with a photocurable liquid polymer. With the platform in upper position, a laser focuses an ultraviolet beam on the upper surface layer, curing that part of the photopolymer to create a solid body. Next, the platform is lowered a bit and the cured polymer is covered with another layer of liquid polymer, after which the sequence is repeated (Kalpakjian 1992). By varying the shape of each new polymer layer, complex geometries can be built up with stereolithography.

23.1.1 Basic Technologies, Characteristics and Fields of Application Today, there exists quite a variety of AM technologies, of which the most important ones are Selective Laser Sintering (SLS) and Selective Laser Melting (SLM), Electronic Beam Melting (EBM), Digital Light Processing (DLP) and Fused Deposition Modeling (FDM). For a detailed exposé of these different technologies, we refer to Gibson et al. (2015); here it suffices to say that they differ widely in the amount and type of materials used, in speed and accuracy, and in their domains of application. In the following, we present a more general overview of AM technologies to lay the foundation for the discussion of supply chain matters later. References given in the further readings section at the end of this chapter may serve as a starting point for a more comprehensive study on the differences of the various AM technologies. Note that the term 3D printing is often used as a synonym for additive manufacturing. In the sequel, we will also use the terms interchangeably since they both refer to layer based manufacturing techniques. Sometimes, the term Additive Manufacturing is used for parts in a production context that need to satisfy tight specifications, whereas 3D printing often refers to the fabrication of less critical consumer products. For most AM technologies, the process starts with a user-defined 3D CAD file of a component or product. Specific AM software then will be used to cut the 3D CAD into slices that are fed into an AM machine to “build-up” the component layer upon layer as shown in Fig. 23.1. An increasing variety of raw materials becomes available, including ceramic powder, metal or even glass, next to polymers. The CAD file may be generated from a design process, but may also result from a 3D scan of an existing object or body. Because of this, design changes are easily incorporated.

23 Additive Manufacturing and Its Impact …

523

Fig. 23.1 The principle of additive manufacturing (EOS 2015)

The thickness of the layers may be in the order of microns; naturally, the thinner the layer, the more detail and accuracy can be achieved. The unique production process of AM technologies has several implications for future manufacturing. Let us therefore look at some potential benefits that are derived from the basic properties of AM technologies. • Design freedom to produce complex and tailored parts The design freedom of AM technologies is certainly one of the main benefits. Design compromises to improve the manufacturability are significantly less limiting with AM technologies than with conventional methods and thus facilitate a more usage-driven design. Complex structures can be built with an optimal balance between strength and material usage that is not feasible with subtractive technologies. Consequences can be observed in the aerospace industry where lightweight designs, that are producible with AM only, lead to significant fuel savings. Figure 23.2 exemplifies this opportunity and shows the conventional (a) and the AM design (b) of a hinge bracket that is used in aircrafts. Overall, the AM-enabled topology optimization lead to a weight reduction of 25%.

Fig. 23.2 Conventional (a) and AM (b) design for a hinge bracket used in aircrafts (Knofius et al. 2017)

524

H. Zijm et al.

Other common examples for design improvements concern heat exchangers or valves where thermal control or flow resistance are improved. However, not only for individual parts AM’s design freedom holds significant implications. Complex parts that previously were assembled from various lower level parts and components, may become producible as an one piece part. Joints or screws that previously were required for the assembly become superfluous. Hence, not only the weight reduces, but potential failure modes like cracks or erosion can be avoided. Supply chain configurations may simplify significantly because the number of distinct parts that need to be sourced, tracked and inspected decreases. We will pay further attention to the implications of the latter aspect in Sect. 23.2.1. • Reduced material waste and energy consumption from an operational perspective The reduction of materials usage is obvious in additive processes. Note that the materials are fed to an AM machine in different states than those used in subtractive processes (e.g. granular or fluid). Combined with the more uniform requirement for raw materials, this characteristic may compensate for the often energy-intensive production process. Indirect effects caused by lower weight or optimized part properties may even further reduce the energy consumption. Hence, from a lifecycle perspective, the energy balance of 3D-printed parts may well turn out to be positive. • High level of customization The possibility to design products according to customer specifications and to manufacture them on-demand, using only basic materials, is entirely due to the fact that both the design and the manufacturing process are highly digitalized. Tooling or product-dependent setup process are (usually) not required, making AM an extremely flexible technology. Thus, for instance, design changes or product changeovers are easily realized while the production process remains unaltered. As we will discuss later, especially for medical and dental applications where customer-specific solutions are paramount, AM transformed entire supply chains. • Faster time to market The fact that design and manufacturing process are so closely intertwined, together with the fact that the product is built up in one piece from raw materials only, eliminated a number of steps in assembled processes. Hence, AM may significantly reduce the time to market which may secure a competitive advantage. Risks associated with market failure decrease, given low setup and tooling costs. Accordingly, it is likely that AM may support a more aggressive market strategy. In addition, fast design changes based on market feedback become possible and thus give rise to more aggressive business strategies. Despite the potential of AM technologies, it would be incorrect to assume that AM is about to replace conventional production methods. It appears more likely that AM technologies complement rather than replace conventional production methods.

23 Additive Manufacturing and Its Impact …

525

The subsequent discussion should clarify today’s shortcomings of AM technologies, and the trade-offs involved when compared to conventional manufacturing methods. Most prominent is the misperception that an AM process alone produces industrial grade parts. Instead, major post-processing steps are often required for achieving sufficient quality. Support structures may need to be removed, or treatments may be required to improve material properties. Furthermore, process variability, inherent to today’s AM methods, often invokes extensive quality controls that increase production lead times and costs. In some fields, like the aerospace sector, process variability complicates the economic certification of safety critical parts. Additional concerns are raised by the digital nature of AM methods. While increasing flexibility, businesses worry about the protection of intellectual property rights and liability. The latter is clarified by a simple example. Consider an innovative company who offers 3D-designs of his products for sale. After a customer printed this product (maybe with slight alterations to the design) and it fails, who is hold responsible for a failure: the company, the printing service provider, the AM equipment manufacturer? Today, no standardized legal agreements are in place, which creates uncertainty about promising new business models. Finally, available materials, size limitations and short development cycles of AM equipment may hamper the breakthrough of AM technologies in several fields. For example, additive manufacturing of electronic parts is still a major challenge. However, it is not only due to these shortcomings when companies decide against AM methods and use conventional manufacturing methods instead. This becomes clear if we compare AM methods with injection molding. Injection molding, which is itself a production technique capable to produce parts in one piece from raw materials, is quite the opposite of AM in terms of flexibility. Based upon a product specification, special molds have to be designed and fabricated at high costs. However, products manufactured with injection molding are relatively cheap for high volumes, and the actual production time of a single product may be a matter of seconds. This makes injection molding suitable for mass production. On the other hand, AM is not suitable at all for mass production for the following reasons. A print of large and complex product geometries may take several hours, while often being highly energy-intensive. In combination with low economies of scale, this causes relatively high marginal production costs. However, small-batch or one-of-a-kind production often profit from the low setup and tooling cost of AM and thus are considered as domains that benefit most from AM technologies. Subsequently, we will detail our discussion on application areas of AM. We will have a look at the rapidly growing number of industries and sectors, typically those where advantages like high customization, light weight and short time to market count most. The strategic research agenda of the Additive Manufacturing Platform (cf. AM Platform 2014) mentions a number of key domains which have adopted AM as a key technology, including: • Medical and dental applications: In particular, titanium alloys have been extensively used as powder material for fabricating orthopedic/orthodontic implants. Other applications can be found in e.g. the hearing aid industry, which has made

526

H. Zijm et al.

Fig. 23.3 Percentage of industrial and public sectors using additive manufacturing (Wohlers 2014)

the transition to almost 100% AM. The key driver for these type of applications is typically the ability to provide customized solutions for an affordable price. • The Aerospace industry, which develops devices based on again titanium and nickel alloys, next to polymers. Today, main business driver are weight and attributed fuel savings. Projects as NASA’s ‘zero gravity’ 3D printer meant to produce spare parts in the International Space Station (ISS) show interest to further expand AM’s application area, cf. TechCrunch (2016). Future scenarios where, for instance, downtime of airplanes are reduced with printed spare parts are likely. • Automotive, which is the second largest sector adopting AM. In this sector, the design flexibility is one of the most important arguments to move to AM, next to the fast realization of prototyped or low-volume car parts. Experimentation with large-scale prints may motivate further applications and indicate the interest to secure weight and thus fuel savings very much like in the aerospace sector, cf. Ford (2017). • Consumer products: to date the largest but also the most diverse sector that has embraced AM technologies. Mostly, the technology is still used for prototyping but a quickly growing number of end-users has discovered AM to deliver highly personalized devices and products, ranging from toys and busts to home furnishings to fashion items (robes, shoes). Other fields of application include industrial machinery, the military domain and architecture (prototyping). An overview of the most important applications is found in Fig. 23.3, based upon Wohlers (2014). In a Harvard Business Review publication, McCue (2015) reports that 30% of the Top 300 largest global enterprises are now using or evaluating the potential of AM. Some of the companies that are already exploiting AM technology are:

23 Additive Manufacturing and Its Impact …

527

General Electric (jet engines, medical devices), Lockheed Martin, Airbus and Boeing (aerospace and defense), and Aurora Flight Sciences (Unmanned Aerial Vehicles), cf. D’Aveni (2015). In summary, AM has quickly gained a firm position in the manufacturing arena. Based upon the observed drop of the cost of AM systems with 50% in the last decade (Thomas 2016) and a further expansion of the materials range, an increased penetration of the manufacturing arena is generally expected. Accordingly, forecasts of Siemens (2014a), a multi-national conglomerate, predict an AM market growth of 300% within the next 10 years.

23.1.2 Case Study: Additive Manufacturing for Customized Mountain Bikes (Based on a case study published by Renishaw (2016)) A mountain bike is a sports bicycle designed for off-road cycling. In principle, they are similar to other bikes but they should withstand heavy use in rough terrain. For that reason, they typically include a front or full suspension, broad tires, durable wheels, powerful brakes and low gear ratios to enable climbing steep hills. High-end mountain bikes often have a frame made from carbonfiber reinforced resins, which are molded. Because of the molding process, the bikes are usually made only in a limited number of sizes whereas for sportsmen and—women it is extremely important to ride a bike that perfectly matches the body’s measures and powers, in other words, that fits the individual rider. Robot Bike Co. is a company that took up the challenge to make customized mountain bikes. To that end, they came up with a new design consisting of a series of carbon fiber-reinforced tubes that needed to be joined by titanium lugs to constitute the frame. The material titanium was chosen for the lugs because of its ability to carry high pressures while carbon fiber-reinforced tubes are both strong and light. The challenge is in the design of the titanium joints. The angles used in the joint, together with the lengths of the sections of tubing, should be unique to match the measures and wishes of the ultimate rider. Robot Bike Co. sought cooperation with a few design companies to come up with an optimal design in terms of weight and material use, subject to stress and strain conditions and matching the individual customer’s need. The result is a bespoke bike, consisting of a number of parts that are uniquely tuned to the specific customer. Advanced CAD and simulation tools were used to produce a parametric CAD engine for mountain bike design. Subsequently, Renishaw, a worldwide operating company with a reputed additive manufacturing center, took up the challenge to manufacture the titanium lugs. The production process starts with bespoke CAD geometries produced from bike design software on the Robot Bike Co. website, where customers can input the required body measurements. The geometries are next imported into Renishaw’s buildpreparation software, where the optimum orientation for each part is selected

528

H. Zijm et al.

Fig. 23.4 Titanium lugs with build plate, produced by additive manufacturing (Renishaw 2016)

and the support structures required for a successful build are added. The eleven lugs required for each bike are grouped together, along with their supports, to be produced in a single build (cf. Fig. 23.4). Upon completion, the build plate with the eleven parts attached is removed from the laser powder bed fusion system and heat treated. Then, the individual parts are separated from the plate and from each other. Some of the lugs require finish machining to produce precision bearing features. The production process is completed with inspection of each part on a co-ordinate measuring machine. The frames assembled from the specialized carbon fiber-reinforced tubes and the titanium joints have passed a variety of tough quality and safety tests. As a result, Robot Bike Co. is now able to deliver high quality bespoke mountain bikes, thanks to the customization abilities of additive manufacturing technology.

23.2 Additive Manufacturing and Supply Chains (Advanced) In this section, we systematically investigate the impact of AM on various aspects of the supply chain. Generally, a supply chain for conventional manufactured products is characterized by a number of functional areas (component manufacturing, final assembly, raw materials and component storage, distribution warehouses, and in general transport between any two facilities). Upon disposal by its user, the product may be returned and possibly its components can be re-used, see Fig. 23.5. For the sake of a fair comparison, we consider the case of a product that is directly delivered from the distribution center to the end-user (no retail functions involved). Next, we systematically evaluate how the main characteristics of the supply chain

23 Additive Manufacturing and Its Impact …

529

Fig. 23.5 Supply chain for a customized product using conventional manufacturing (RM  Raw Materials)

will change, once the assembled product is replaced by a product manufactured using AM. We concentrate on typical operational characteristics, such as lead times, in process inventories, system responsiveness and supply chains costs.

23.2.1 Impact of AM on the Supply Chain Already earlier, we noted that the design freedom that product developers experience is an attractive feature of AM technologies. At the same time, the manufacture of products is generally perceived to be quite costly, due to, for example, the high energy use of the AM process and the relatively long production time, cf. Sect. 23.1. But what we will see is that the consolidation of component functionalities into one piece in combination with decentralized manufacturing may compensate for these disadvantages. Here, with consolidation we indicate the redesign of an assembled part or product, such that its constituting components are replaced by fewer, but more complex components, which hence integrate various functionalities. Supply Chain Structure Typically, in conventional manufacturing all components need to be available before assembly can start. If a number of components is skipped because their functionality is integrated in a 3D-printed part, the corresponding supply relations are no longer needed which reduces synchronization efforts. Consequently, the need for multi-level production planning decreases and thus lessens associated risks and costs. Supply Chain Inventories Consolidation of functions due to integration implies that raw materials are directly used in final product manufacturing. Consequently, downstream stocks are no longer

530

H. Zijm et al.

needed. In particular for customized products, this may have far-reaching consequences. Today, customization is typically achieved with modular product designs, where customer requirements are satisfied by combining the right components in one final product. While a modular product design enables postponement of a desired product configuration, modularity comes with the sacrifice of often high inventories of many different components to achieve fast response times. With AM, component inventories may become (partially) avoidable due to the strong digitalization of the manufacturing process, cf. Sect. 23.1. Finally, raw material inventories reduce due to the fact that less material is wasted compared to subtractive manufacturing techniques. Decentralized Manufacturing and Transport Times 3D printing facilities can be rather modest and therefore lend themselves to decentralized manufacturing, even more if component stocks are limited (in case of many component stocks, centralized inventories that allow for pooling of stocks are more natural). Clearly, production close to customers may reduce final distribution and transportation time considerably. The fact that raw materials have to be supplied to these decentralized locations is not a serious problem since it concerns bulk transport of only a few basic materials. Growing numbers of printing farms, where service providers offer printing capacity on demand and location, exemplify the perceived value of this business model. The need for spare parts at remote locations as for example often encountered in the mining industry or during humanitarian missions may profit from decentralized manufacturing as well. Component Lead Times Most components in a classical assembly process are ordered from external suppliers, but even if they are manufactured in a department of the same factory that delivers the final product, they have to be planned in advance, according to e.g. an MRP methodology (cf. Chap. 12). Typically, components are produced in batches for reasons of efficiency, blowing up lead times and inventories even further, while additional delay may occur due to queueing problems in case of limited machine capacity. Hence, consolidation using AM may lead to severe lead-time reductions in the overall supply chain. Customer Responsiveness As mentioned earlier, in conventional manufacturing short customer response times are generally realized via high inventories of either final products or modules and components. Commonly this approach causes a significant amount of capital invested and hence diminish a company’s liquidity. The opposite approach are less stocks but longer customer waiting times since specific components need to be ordered, which may reduce sales. The flexibility and speed to setup or alter a production process with AM may provide an alternative. In particular, often investment in raw material stocks suffices in order to provide the required responsiveness. This implies a shift of the Customer Order Decoupling Point (or: push-pull boundary) upstream the supply chain (cf. Chap. 5). Reduced transport times, achieved by customer’s proximity, also contribute to short customer response times.

23 Additive Manufacturing and Its Impact …

531

23.2.2 Costs and Benefits of AM in the Supply Chain As becomes clear from the discussion in the previous section, the ultimate decision to invest in AM depends on a careful trade-off between costs and benefits. AM may appear as an expensive process, but when expanding the view to its impact on a supply chain level, there are significant benefits in terms of inventory reduction (hence enhancing liquidity, i.e. freeing capital for investments), shorter lead and customer response times. From an sustainability point of view, the resulting picture is less clear; reduction of materials waste during production is obviously beneficial. On the contrary, the value of the product after discarding it is at least debatable, whereas for classically manufactured products, it may be feasible to reuse functioning components. AM is a manufacturing technology that highly depends on digitalization, and therefore requires skills that are quite different from classical manufacturing. Typically, designers and software engineers are highly trained specialists, whereas the required amount of lower skilled labor at a machine operator level is less than in conventional manufacturing. As a result, the cost of manual labor as a percentage of the overall product price is rather low, which is an argument to locate the production processes in high wage countries, instead of transferring it to low wage countries. Indeed, observing that labor rates in China are now about five times as high as in 1990 (EPRS 2014), quite many companies have decided to reshore the manufacturing of high-tech products back to Europe and the US. In addition, as we saw, decentralized production becomes attractive once a major part of the process is primarily software-based. For the same reason, AM appears to be an excellent candidate for cloud manufacturing (cf. Chap. 12). Although the price of AM devices is sharply decreasing, the costs of high level professional equipment is still significant, in particular in view of the fact that most production concerns slow moving items while technology life cycles of AM equipment are short. However, the combination of product design flexibility, digital transfer of product data and the relatively stable (less varying) process characteristics makes it relatively simple to outsource the production process to specialized AM service providers. Indeed, the number of such providers is rapidly growing; they may offer not only manufacturing processes as such, but more and more also design support. There is no doubt that the combination of AM and cloud manufacturing offers a fascinating outlook. In the next section, we shift our focus to the implications of AM for the aftersales service logistics of expensive capital goods. In particular, we discuss how the associated low-volume, high value spare part business may change with the emergence of AM technologies.

532

H. Zijm et al.

23.3 Additive Manufacturing of Spare Parts (State-of-the-Art) So far, we have discussed products intended to end users primarily. An important additional group are products that serve as parts in larger assets (machines, vehicles, aircraft or building constructions). Such parts may eventually malfunction or even completely fail, and then need to be either repaired or replaced. Since the costs of downtime of a capital asset may be high, often one chooses to limit downtime by replacing the malfunctioning part, and either disposing it or repairing it off-line. In any case, there should be spare parts available. As a result, one may observe high spare parts inventories (of many different items) to anticipate any possible failure, representing a significant investment. Typically, demand for spare parts is low (slow movers), as failures should not occur often for a well-designed product. Hence, a lot of stock is actually tied up for low demand items.

23.3.1 Specific Opportunities of AM in After-Sales Service Supply Chains As discussed by Gibson et al. (2015), low volume spare parts are mostly produced on generic (i.e. non-dedicated) equipment (e.g. CNC workstations). The fact that they require a set-up time in addition to special tools and fixtures make their overall manufacturing costs relatively high. It seems then natural to think about 3D-printing these parts on demand as an alternative. Very much as for regular parts, it is likely that, as a consequence of 3D-printing, inventories decrease while the responsiveness can be maintained or even improved. Given the long turnaround times of slow moving parts, it is instructive to note that low inventories simultaneously reduce the risk of obsolescence, i.e. storing parts that in the end are not used at all anymore. However, there are more (unique) reasons why AM may qualify as great fit for the spare part business. Upstream in the supply chain it is not uncommon that at some point in time a supplier decides to discontinue the production of spare parts, for example because the low volumes are not commercially interesting anymore when compared to other opportunities. At best, the asset owner (or service provider) gets a chance to once purchase a final set of parts to cover possible demand during the entire remaining lifetime of its assets (cf. Behfard et al. 2015), but in any case it is clear that such a supply risk may in the short run cause severe problems. In such a case, the possibility to print the part directly from raw materials, based upon a stored CAD file, presents a welcome alternative. Also relevant for the spare parts business might be the option to repair parts with AM technology. Worn out parts that were previously discarded or expensive to repair may become repairable with AM and thus significantly reduce the maintenance cost of the capital good. The potential is demonstrated by means of a burner tip used in gas turbines at Siemens. Siemens (2014b) was able to reduce the repair lead time by

23 Additive Manufacturing and Its Impact …

533

Fig. 23.6 Repairing burner tip with additive manufacturing (Andersson et al. 2016)

90% and the associated repair cost by 30%. Figure 23.6 illustrates how a new burner tip is printed and afterwards attached to the burner. Even if a conventionally manufactured spare part appears to be the technically and commercially best solution, it still may appear valuable to 3D-print a spare part as a temporary fix. That is, the printed part would bridge the period until the intended replaceable becomes available. Interesting for this application is that a temporary fix might even turn-out valuable if it yields lower performance rates. Preliminary results indicate that, in case the failure rate is ten times higher than for the conventional manufactured part, a temporary fix still may remain profitable (Knofius et al. 2017). There are also arguments against the use of printed spare parts. Part characteristics may worsen significantly. For example, based on the process characteristics of AM, often unit cost and reliability of the AM part are inferior compared to the conventionally manufactured version. In addition, conventional parts that are built up from components may be repaired by replacing only a malfunctioning component after which the part is re-assembled. A 3D-printed part on the other hand may have to be discarded as a whole, resulting in more waste while its replacement by a complete new part is definitely more costly than the replacement of a simple malfunctioning component or subcomponent in a conventionally manufactured part. Similarly, possible commonality effects are lost because of the higher customization of the consolidated part (Knofius et al. 2016b). Overall, it should be clear that also for the spare part business a detailed assessment of the costs and benefits of applying AM are necessary. It appears certain, however, that AM technologies, with their unique characteristics, hold the opportunity to transform several fields of the spare part business in the near future. The described applications may lend themselves as valuable starting point to motivate the economic value. Subsequently, we discuss a more algorithmic approach for the selection of spare parts having item and supply chain characteristics suitable for AM as mentioned before.

534

H. Zijm et al.

23.3.2 Selecting Spare Parts for 3D Printing As explained in the previous section, spare parts used in capital assets seem to be potential candidates for AM, based on their relative low demand, their often high manufacturing costs, and their short customer order lead times. For many practitioners however, it is not clear how to identify the most promising spare parts. Overall, asset owners and service providers often deal with spare parts assortments easily exceeding hundreds of thousands of spare parts. Amid these circumstances, a manual assessment of each spare part appears virtually impossible. In this section, we describe a method that prioritizes spare parts according to their likelihood of being valuable for AM. As we will explain, the method does not require any manual assessment, thus providing a systematic approach to prioritize and therefore afterwards to evaluate the most promising spare parts first. The ranking method is based on information easily retrievable from company databases that describe the spare part characteristics and give insights about the value of printing these parts. For instance, in line with the explanations in the previous sections, a high demand rate may indicate that conventional manufacturing methods are a good fit and AM is less likely to be interesting. On the other hand, if the demand rate is low, it might be worthwhile to assess the potential benefit of producing the spare part with AM. Furthermore, we take into account company goals that give insight about how AM offers the highest value to the specific business. That is, companies focusing on cost reductions may profit differently from AM than companies having a strong quality focus. As listed below, the method consist out of four steps that we will describe in the next subsections. 1. Determining the spare part assortment to be investigated 2. Assigning a score to each spare part attribute, based on a comparison with the same attributes of other spare parts (horizontal comparison) 3. Assigning weights to spare part attributes, specifying the degree to which these attributes contribute to overall company goals (vertical comparison) 4. Determining a weighted score for each part For a more elaborate discussion of the methodology, we refer to Knofius et al. (2016a). Determining the Spare Part Assortment to Be Investigated To determine what spare parts in principle might be eligible to be produced by AM, we need to specify what are major determinants. To that end, we characterize each part by two physical and eight operational attributes, as specified in Table 23.1. Note that the attribute agreed response time refers to the maximum time between receipt of a customer request for a part and the supply of a ready-for-use part at the customer site. The attribute remaining usage period describes the estimated time

23 Additive Manufacturing and Its Impact …

535

Table 23.1 Spare parts characterized by physical and operational attributes Parts characterization by attributes Physical attributes

Material type (electronic, metal, plastic) Part size (dm3 )

Operational attributes

Demand rate (estimated number of requests per month) Resupply lead time (in months) Agreed response time (in months) Remaining usage period (in months) Manufacturing/order costs (in units of EURO 10.000) Safety stock costs (based on 2% of manufacturing costs, per month) Number of supply options (number of potential suppliers) Supply risk (probability of supply discontinuation in the next year)

horizon the assets, where the spare part is used for, is still used/supported by the company. Finally, the attribute supply risk describes the likelihood that supply for the spare part is discontinued in the near future. As exemplified later supply risk is measurable with methods as for instance proposed by Jaarsveld and Dekker (2011). Empirically, the ten attributes proved to be suitable and often retrievable form company data. It needs to be noted however, that occasionally not exactly the information is available but instead some sort of substitute. Also, it may occur that information for different spare part groups may reside with different partners (OEM’s, asset owners, service providers). Depending on the records for each spare part group, it may become necessary to separate the analysis in order to maintain the comparability between all spare parts. Similarly, if for certain spare parts significantly less information is available compared to the overall population, it may appear necessary to exclude them from the analysis. Consequently, we usually end up with a subset of the entire spare part population. In Table 23.2, we present a snapshot of the resulting assortment list.

Table 23.2 Details of assortment output list Part ID 1 Material (electronic, metal, plastic)

E

2

3

···

M

P

···

Part size (dm3 )

1

3

4

···

Remaining usage period (in months)

21

56

12

···

Manufacturing/order costs

5

15

1

···

Number of supply options

1

14

3

···

Supply risk

0.21

0.50

0.35

···

···

···

···

···

···

536

H. Zijm et al.

Assigning Scores to Spare Part Attributes For the physical attributes, a binary score (0–1) is applied, indicating whether a physical attribute value allows for AM anyhow, based on technological feasibility. Hence, a zero score on any physical attribute rules out the particular spare part for further investigation. For instance, based on current technology standards, a part size exceeding 125 dm3 is rather unlikely to be economically printable; the same holds for electronics. Nevertheless, we would like to take this opportunity to emphasize that if both physical attributes are evaluated to 1, it does not guarantee technological feasibility. It merely increases the probability of technological feasibility within the ranking. Later, after the ranking has been generated, an in-depth study of part attributes, not as easy to retrievable as material and part size, is necessary to draw a final conclusion about the technological feasibility. For the operational attributes, we translate the value of each attribute into a linear score, ranging from 0 to 1 (needed since the operational attributes are defined in various dimensions). The part attribute score indicates how a part scores with respect to the economic potential of AM, when compared to other parts on solely this particular attribute. Hence, a score of 1 on a particular part attribute means that this part scores better (or not worse) than any other part on this particular attribute. A score of 0 means that all other parts score better on this particular attribute. Assigning Weights to Attributes For the assignment of weights to part attributes, the primary consideration is based on how the improvement of a part’s attribute contributes to an improvement of operational performance, or even more, to strategic company goals. Regarding the latter, a common distinction is often made between companies focusing on costs reduction and companies focusing on operational improvements, while sometimes a third consideration concerns resilience against possible disruptions (see e.g. Chopra and Meindl 2016, for a similar classification). In the context of spare parts management, we have defined improvements of three cost-related criteria, three operational performance criteria and one resilience-related criterion, see Table 23.3. In this table, we also indicated what a change of a particular part attribute might mean in terms of these criteria improvements, and hence, what it would contribute to the overall company goal. The table should be read as follows: if a particular attribute x (left column) takes the relative value y (central area), then the alternative of producing the part with AM might help to improve on criterion z (upper row). As an example: if the resupply lead time of an existing part is long, then the alternative of AM might help to reduce safety stocks, improve the responsiveness of the supply chain, may add to the possibility to postpone customized production and finally also support the application of a temporary fix. For a more extensive justification of the impact of changing a certain spare part attribute, we refer to Knofius et al. (2016a). In general, of course, a company will almost never entirely classify as cost-driven or operations-driven or resilience-driven, so it is important to assign weights to each of these three company objectives (see the colors in Table 23.3). To this end, we apply the Analytical Hierarchy Process (AHP) in two subsequent stages (AHP, cf. Saaty

23 Additive Manufacturing and Its Impact …

537

Table 23.3 Value range of spare part attributes that indicate improvement potential when switching to 3D printing

2008). In the first stage, company representatives compare any two company goals, resulting in an overall weight of each of the three company goals (weighs sum up to one, hence each weight indicates the relative importance of the associated goal). Due to the pairwise comparison, inconsistency becomes controllable and decision complexity is prevented. From a small aggregation of Table 23.3, we learn that seven attributes relate to potential cost reductions, six to potential operational improvement, and two to supply risk reduction. Note that the two physical attributes are not related to any company goal. They concern technological feasibility of AM and hence are independent of these goals. In the second stage of the AHP process, pairwise comparisons between any two assigned spare part attributes are made by company representatives again, resulting in a relative importance of each attribute in contributing to each company goal when switching to AM. The partial weight (partial, since it relates to a specific company goal) is now determined by multiplying the relative importance of each company goal with the relative importance of each attribute related to that company goal. Figure 23.7 presents a simplified example of the calculation, based on the four attributes listed in Table 23.2. Finally, the weight of an attribute is obtained by summing up its partial weights. For example: the weight of attribute Supply risks is equal to 0.104 + 0.218  0.322. It is important to realize that the attribute weights just indicate the relative importance of each attribute to the overall company goals when switching to AM.

538

H. Zijm et al.

Fig. 23.7 Calculation of partial weights of four attributes (blue digits)

Determining a Weighted Score for Each Spare Part Based upon the part attributes scores and the weights of the attributes, we are finally able to calculate a total score for each spare part, using the following calculation scheme. For each part, a. Determine a weighted score per operational attribute, by multiplying the weight with the score. b. Multiply the scores of the physical attributes, called PAS, which results in either a 0 or a 1. c. Determine the sum of the weighted scores of the operational attributes, called OAS. d. Multiply PAS with OAS to determine the overall score of the spare part. Note that if only one physical attribute has a score of zero, the overall parts score is zero as well (AM of this part appears technological infeasible). Table 23.4 provides an example of the calculation, based on the four operational attributes used in Table 23.2 and Fig. 23.7. This concludes the description of the spare parts ranking procedure. Based on the resulting ranking, the analysis would proceed with a manual assessment starting from the best scoring part. In that way, it is more certain that most promising spare parts are assessed first and the risk of disregarding interesting parts is minimized. In the following case study, we show how the method proved its value in spare parts management in the aviation industry.

23 Additive Manufacturing and Its Impact …

539

Table 23.4 Calculating the overall parts score as a candidate for additive manufacturing Attribute Value Weight Score Weighted score Material type

Metal

_

1

1

Part size Supply risk

0.5 20

_ 0.322

1 0.21

1 0.06762

Remaining usage period

15

0.074

0.31

0.02294

Supply options

5

0.460

0.48

0.2208

Manufacturing/order costs

48

0.144

0.24

0.03456

Overall score

0.34592

23.3.3 Case Study: Selecting Spare Parts for AM in the Aviation Industry The selection procedure presented in the preceding section has been tested during a field study at a part supplier in the aviation industry, with more than 400,000 spare parts. After an evaluation of the data availability and data cleaning, we decided together with company representatives to base the analysis on 40.330 spare parts. Additionally, this selection ensured design ownership and excluded confidential parts. As basis for the analysis, we used the two physical and eight operational attributes as specified in Sect. 23.3.1. Given that we did not find a suitable data source for the agreed response time, we had to drop this attribute. Furthermore, we used the part number as a substitute for the part size, because direct information about the part size was often not accessible with data base queries. Fortunately, the company-specific numbering system relates part size to the part number and thus turned out to be a good proxy. In addition, we used the attribute airplane type instead of number of supply options. The reason was that for spare parts that are exclusively used in a specific airplane type, demand is fulfilled by dismantling phased-out airplanes. Other information about the number of supply options was not easily retrievable. Finally, we substituted supply risk by the attribute survival probability where the survival probability defines the chance that a spare part supplier will be available within one year. This measure appeared to be available in this company for most analyzed spare parts and was computed based on the method of Jaarsveld and Dekker (2011). In Table 23.5, we present an overview of all used attributes associated with the weights derived from the AHP method. During the analysis, we found that 34.140 spare parts did not pass one of the physical attribute tests, hence were not feasible to print from a technological perspective (at least with the current state of technology). The remaining 6.190

540

H. Zijm et al.

Table 23.5 Attributes and weights used in the aviation spare parts study Attribute Weight Explanation Part number



The part number gives insights about spare part size

Material type



Safety stock costs

0.18

Indicates the material type e.g. electronic, composite or metal High safety stock may be reduced with AM

Manufacturing/order costs 0.17

High sourcing costs may be reduced with AM technology

Demand rate

0.16

Survival probability

0.13

Remaining usage period

0.13

For low volume production AM may reduce order costs Spare parts with high supply risk could be obtained with AM An early lifecycle phase may indicate high saving potentials of operational costs

Resupply lead time

0.13

AM may reduce long resupply lead time and thus decrease safety stocks

Airplane type

0.10

Specific airplanes obtain less spare parts from dismantling

spare parts were ranked, which resulted in a score distribution as shown in Fig. 23.8. Based on the ranking, the case company could already identify 1,141 technologically feasible and economically beneficial business cases. Today, only the non-formalized certification process holdback the changeover to AM for some safety-critical parts. In the near future however, this is expected to change. A typical example is a fitting stud used for the attachment of a safety belt as illustrated in Fig. 23.9. For this case, it is estimated that the resupply lead-time might be reduced by some 40% and the order costs by about 70% with AM. The prospect of this improvement potential stimulated a reengineering project for the fitting stud despite high efforts for certification. This outcome demonstrates the benefit of the developed top-down approach: practitioners probably would have disregarded the fitting stud due to its high certification effort. In comparison, the ranking method typically exposes promising characteristics for high-scoring items and thus justifies an assessment of the part in more detail.

Fig. 23.8 Score distribution of parts in aviation field study

Percentage of spare part assortment

23 Additive Manufacturing and Its Impact …

541 41% 35% 16% 6%

2% (0;0.2)

(0.2;0.4)

(0.4;0.6)

(0.6;0.8)

(0.8;0.1)

Scores

Fig. 23.9 Safety belt with fitting stud (grey)

23.4 Further Reading A good introduction in the development of AM technologies and their strategic impact is given in the Additive Manufacturing Strategic Research Agenda 2014, cf. AM Platform (2014) and in Deloitte (2014), see also Wohlers (2014) and subsequent reports of the Wohlers Association. An introduction to AM technologies is provided by Gibson et al. (2015). The scientific literature on AM and more in particular on its impact on the supply chain is still relatively young, see e.g. Thomas (2016), Khajavi et al. (2014), Sasson and Johnson (2016), Oettmeier and Hofmann (2016) and Walter et al. (2004). Sirichakwal and Conner (2016) and Holmström et al. (2010) discuss the impact of AM on spare parts inventories in the supply chain. Parts selection is discussed by the current authors in Knofius et al. (2016a). Consolidation aspects of additive manufacturing are discussed in Yang et al. (2015) and Knofius et al. (2016b).

References AM Platform (2014) Additive manufacturing: strategic research agenda 2014 Andersson O, Graichen A, Brodin H, Navrotsky V (2016) Developing additive manufacturing technology for burner repair. ASME J Eng Gas Turbines Power 139(3):031506 (9 pp). https://do i.org/10.1115/1.4034235 Behfard S, van der Heijden M, Al Hanbali A, Zijm WHM (2015) Last time buy and repair decisions for spare parts. Eur J Oper Res 498–510

542

H. Zijm et al.

Chopra S, Meindl P (2016) Supply chain management: strategy, planning and operation, 6th edn. Pearson Education Limited, Essex D’Aveni R (2015) The 3-D printing revolution. Harvard Business Review, May 2015. https://hbr.o rg/2015/05/the-3-d-printing-revolution Deloitte (2014) 3D opportunity in areospace and defense: additive manufacturing takes flight. Deloitte University Press EOS (2015) Additive manufacturing, laser-sintering and industrial 3d printing—benefits and functional principle. https://www.eos.info/additive_manufacturing/for_technology_interested EPRS (2014) Reshoring of EU manufacturing, briefing to the EU Parliament by the European Parliamentary Research Service 21/03/2014 Ford (2017) Ford tests large-scale 3D printing with light-weighting and personalization in mind. https://media.ford.com/content/fordmedia/fna/us/en/news/2017/03/06/ford-tests-large-sc ale-3d-printing.html. Accessed 03 July 2017 Gibson I, Rosen D, Stucker B (2015) Additive manufacturing technologies: 3D printing, rapid prototyping and direct digital manufacturing, 2nd edn. Springer, New York Holmström J, Partanen J, Tuomi J, Walter M (2010) Rapid manufacturing in the spare parts supply chain: alternative approaches to capacity deployment. J Manuf Technol Manag 21(6):687–697 Jaarsveld and Dekker (2011) On the use of installed base information for spare parts logistics: a review of ideas and industry practice. Int J Prod Econ 143(2):536–545 Kalpakjian S (1992) Manufacturing engineering and technology, 2nd edn. Addison-Wesley, Reading, MA Khajavi S, Partanen J, Holmström J (2014) Additive manufacturing in the spare parts supply chain. Comput Ind 65(1):50–63 Knofius N, van der Heijden MC, Zijm WHM (2016a) Selecting parts for additive manufacturing in service logistics. J Manuf Technol Manag 27(7):915–931 Knofius N, van der Heijden MC, Zijm WHM (2016b) Consolidating spare parts for asset maintenance with additive manufacturing. Research Report, University of Twente (submitted for publication) Knofius N, van der Heijden MC, Sleptchenko A, Zijm WHM (2017) Improving effectiveness of spare part supply by additive manufacturing as dual sourcing option. Research Report, University of Twente (submitted for publication) McCue TJ (2015) 3D printing is changing the way we think. Harvard Business Review, July 2015. https://hbr.org/2015/07/3d-printing-is-changing-the-way-we-think&cm_sp=Article-_-Lin ks-_-Top%20of%20Page%20Recirculation Oettmeier K, Hofmann E (2016) Impact of additive manufacturing technology adoption on supply chain management processes and components. J Manuf Technol Manag 27(7):944–968 Renishaw (2016) Metal additive manufacturing technology used for customised mountain bikes. http://www.renishaw.nl/nl/metal-additive-manufacturing-technology-used-for-customise d-mountain-bikes–39741 Saaty T (2008) Decision making with the analytic hierarchy process. Int J Serv Sci 1(1):83–98 Sasson A, Johnson J (2016) The 3D printing order: variability, supercenters and supply chain reconfigurations. Int J Phys Distrib Logist Manag 46(1):82–94 Siemens (2014a) Additive manufacturing: 3D printing: facts & forecasts. https://www.siemens.c om/innovation/en/home/pictures-of-the-future/industry-and-automation/Additive-manufacturin g-facts-and-forecasts.html. Accessed 03 July 2017 Siemens (2014b) Additive manufacturing: from powders to finished products. https://www.siemen s.com/innovation/en/home/pictures-of-the-future/industry-and-automation/additive-manufactur ing-from-powders-to-finished-products.html. Accessed 03 July 2017 Sirichakwal I, Conner B (2016) Implications of additive manufacturing for spare parts inventory. 3D Print Addit Manuf 3(1):56–63 TechCrunch (2016) NASA is sending a 3D printer to space that you can use. https://techcrunch.co m/2016/03/19/nasa-is-sending-a-3d-printer-to-space-that-you-can-use/

23 Additive Manufacturing and Its Impact …

543

Thomas D (2016) Costs, benefits, and adoption of additive manufacturing: a supply chain perspective. Int J Adv Manuf Technol 85(5):1857–1876 Walter M, Holström J, Yijölä H (2004) Rapid manufacturing and its impact on supply chain management. In: Logistics research network annual conference, Dublin Wohlers (2014) Additive manufacturing and 3-D printing state of the industry. Annual worldwide progress report, 2012 ed. Wohlers Associates, Colorado, USA Yang S, Tang Y, Zhao Y (2015) A new part consolidation method to embrace the design freedom of additive manufacturing. J Manuf Process 20(3):444–449

Chapter 24

Future Technologies in Intralogistics and Material Handling Kai Furmans, Zäzilia Seibold and Andreas Trenkle

Abstract This chapter describes future technologies in Intralogistics and Material Handling. Starting from a description of today’s material handling systems (including two case studies) and an analysis of their shortcomings we derive desirable properties for future material handling systems (basic section). The necessary functions for these systems are explained and samples of modern material handling systems are presented which at least partially implement these properties (advanced section). The state of the art of the challenging functions is explained and references for further reading are given.

24.1 Structure of Today’s Material Handling Systems and Opportunities for Their Improvement (Basic) In order to motivate the desirable properties of design patterns for future Material Handling Systems, we first describe challenges experienced when using today’s systems.

24.1.1 Today’s Material Handling Systems The structure of today’s material handling systems is still largely determined by the developments of automation technologies ranging from the 1970s up to today. These systems have been developed from purely manual operation through mechanized operation to automated systems. The first automated material handling systems were built in the 1970s when the first mainframe computers became commercially available for industry use. Especially the advantages of dynamic storage allocation, which K. Furmans (B) · Z. Seibold · A. Trenkle Institute of Material Handling and Logistics, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_24

545

546

K. Furmans et al.

required fewer storage locations than fixed allocation per part number, was driving the usage of computers for warehouse management. Based on the necessary functions for tracking inventory on storage locations, in a later stage support for order picking, put-away, consolidation of orders, shipping and receiving was added. Automated subsystems like automated storage and retrieval systems (AS/RS), conveyors or Automated Guided Vehicles (AGV) usually bring their own control systems. Typically, pallets or bins have to be identified again, when they are moved from one subsystem to another one. When pallets are moved from the receiving area to the automated warehouse, they have to be identified again at the entrance of the warehouse, either by reading a barcode and retrieving the associated information from a different system or by entering the necessary data like size, weight, product information and manufacturing date manually in the warehouse management system. The different subsystems are necessary, since they control the underlying PLCs (Programmable Logic Controllers) close to real time, which in turn control the actuators and sensors that are directly attached to the material handling systems. The choice for a multilevel design is motivated by two facts. First, computing power used to be expensive and limited. Therefore a trade-off between scope (how many processes are managed by one system) on the one hand and level of detail as well as real time capability on the other hand had to be made. Second, communication is organized in a hierarchical way, which helps to create an aggregated system overview on the respective higher level. Several institutions have standardized multilevel control architectures for material handling systems. The standards describe the tasks to be performed on each level and the communication protocols between the levels. Typically, within a company using automated material handling systems, there are between 5 and 8 levels of systems interacting with each other (see Fig. 24.1). Over time, the subsystems and their components have become more and more standardized and modular. The software, like warehouse management software, material handling coordinators and manufacturing process control have seen a continuous development. In addition, software companies have emerged that are independent of hardware suppliers. However, the combination and coordination of all these software systems requires a careful design, engineering and testing, to successfully route a customer, production or supplier order through the system. This in turn requires a thorough knowledge of the physical layout, the wiring of all the actuators and sensors as well as their respective connection within the switchboards. The interpretation of sensor-signals and their translation towards a state of the material handling system requires comprehensive knowledge about the structure and the inner organization of the material handling system and the links between physical connections and location of sensors and actuators on the one hand and the logical representation in the software system on the other hand.

24 Future Technologies in Intralogistics and Material Handling

547

Fig. 24.1 Multilevel control architecture (Adapted from Arnold/Furmans 2009), ERP Enterprise Resource Planning; MES Manufacturing Execution System; NC Numerical Control; PLC Programmable Logic Controller; AGV Autonomous Guided Vehicle; AS/RS Automated Storage and Retrieval System; OMR Overhead Monorail System

Fig. 24.2 Divert with three directions (Adapted from Arnold/Furmans 2009)

Case Study 1: Decision Making on Machine Control Level In order to illustrate the decision-making process on the lower levels, consider Fig. 24.2. The procedure to be described may slightly differ from one implementation to another one, but serves to give an impression about the complexity of the software required for the coordination of a little task within a material

548

K. Furmans et al.

handling system. The task is to convey a pallet (flow λ) entering at the left side towards the upper right exit (creating the flow λ1 ). To move the pallet, the drives on the roller conveyor are operated, until light barrier (1) is interrupted. Light barrier (1) might also be coupled with a barcode reader or RFID-reader, which is used to identify the pallet or its destination (see below). The local PLC then possibly stops the drive on the roller conveyor and checks the correct positioning of the transfer carriage with respect to the roller conveyor, either by using additional light barriers or a distance measuring system for the transfer carriage. If the correct position is ensured, the roller drives of the roller conveyor as well as those on the transfer carriage are activated and light barrier (2) is monitored. If light barrier (2) is interrupted, the conveyor on the transfer carriage remains switched on until light barriers (1) and (2) are clear. Careful parametrization makes sure that the remaining travel of the pallet is still less than the distance to the edge of the transfer carriage. This is checked by means of light barrier (3). If clear, the drives of the transfer carriage move it towards the destination position. The destination is usually either derived from an address that may be barcoded or contained in an RFID-tag on the pallet. Frequently the code or tag on the pallet only contains an ID for the pallet. In this case, the local PLC determines the destination for the pallet based on information previously obtained from a higher-level material handling coordinator or warehouse management system or requests this information from the upper level system. When the transfer carriage is approaching the destination position, the PLC must make sure that the transfer carriage is positioned sufficiently accurate for the handover of the pallet at the uppermost position of the receiving roller conveyors to the right. This usually requires additional light barriers that are used to check the exact positioning and control of the drives. Here as well, the link between the physical light barriers, the port on a PLC and the logic entity in the software running on the PLC must be created by a close cooperation between the people doing the electrical installation and those who write or configure the software. When the proper alignment of transfer carriage and roller conveyor is established, the roller actuators on the transfer carriage and on the fixed installation on the upper right must be synchronously accelerated in order to move the pallet safely from the transfer carriage onto the roller conveyor which brings it closer to its destination. When the light barriers (3) and (4) have been first interrupted and then cleared again, we can be sure, that the pallet has cleared the transfer carriage and the transfer carriage can return to the starting position, ready to receive another pallet from the incoming roller conveyor. The positioning here requires again the communication with the positioning sensors of the transfer carriage. For a better accuracy of the positioning it might be necessary to adapt the speed of the drives. When travelling a long distance, drive speed might be larger than when close to the end position. The crawl speed, which frequently is used close to the destination position, allows for higher positioning accuracy.

24 Future Technologies in Intralogistics and Material Handling

549

The reaction speed of the sensors, the PLC and the actuators translates into a smaller inaccuracy with respect to the alignment of transfer carriage and roller conveyors, therefore reducing the need for oversizing the dimensions of the conveyor vs. the size of the pallets. If varying speeds are to be used, the PLC must control the distance travelled and the speed of the drives more accurately. Depending on the drives, the speed setting has to be done by the PLC or, if intelligent actuators are used, can be left to the latter, which reduces the complexity of the PLC-programming. As already stated, there are different ways to handle the transportation process, but the resulting complexity remains the same. The level of detail of the previous description is not very high because we only wanted to give an impression why the installation and especially a possible reconfiguration of such a system requires considerable time and skills to implement. Case Study 2: Decision making on process control level The interaction of the material flow coordinator level and the machine control level will be illustrated using the example depicted in Fig. 24.3. Imagine a manufacturing plant with material handling systems for a raw materials warehouse, parts transportation within production, parts transportation from the warehouse or the production buffer to the assembly line, transportation of finished goods to the finished goods warehouse and the shipping area. An ERP-system is used to manage the entire site on an aggregated level. In addition, there is a warehouse management system (or two) for both warehouses, manufacturing process control for the production and assembly line, a material handling coordinator for the coordination of the material handling systems and very likely a PLC in the warehouses (per AS/RS-machine), the AGVs, the overhead monorail system (OMR) and possibly a forklift management system in place. If raw material is ordered for an assembly order at the assembly line, a transportation order is issued, which leads to a retrieval in the raw materials warehouse. The warehouse management system also issues a transport order to any conveyor in the warehouse to move the pallet with the parts to the loading station of the OMR. Then the transport order is passed on to the material handling coordinator, which then makes sure, that the PLC controlling the OMR picks up the right pallet and moves it to the unloading station at the destination. Then a person picks it up and supplies it to the assembly line. The transport order then has to be reported as completed and the parts must be transferred from the raw material inventory to the manufacturing inventory. This might also apply to the pallet itself, if these are inventory-tracked. Five levels of software need to be coordinated to perform the retrieval and the transport of a pallet from the raw material warehouse to the assembly line. This requires:

550

K. Furmans et al.

• The transport order must be known in all subsystems. • If the transport order is changed or cancelled, this must be executed in all subsystems. Since this is rather challenging, cancellations and changes are not allowed in most systems, and a new order is issued to transport the goods back to the origin. • Identification of the pallet and link to the transport order in all subsystems. • A unified method to identify the ultimate destination and thereof deduce the pick-up point and hand-over point of the transport in each subsystem (where does the OMR for instance pick up the pallet and which unloading station is closest to the supply point?). • An update of the status of the transportation order, to be fed to all system levels. • Interfaces between the ERP-level and the coordination level systems as well as between the coordination levels and subsystems which have to be designed, implemented, continuously supervised, and updated when process or software changes are necessary. • If a system is reconfigured (for instance because of introducing a new destination), the master data has to be consistently updated through all system levels.

Shortcomings of structure of today’s material handling systems Productivity, reliability and quality of material handling processes executed by automated systems have increased tremendously over the last thirty years of continuous development of the standards, the subsystems and the communication systems. However, it has become clear, that the multilevel architecture has reached its limits, when flexible, easily adaptable systems are required. A quick and easy rearrangement of today’s Material Handling systems is hampered by the following factors (as shown before): • The decision-making process is usually distributed over several system levels, the local PLCs, the material flow controller, the Warehouse Management system and the ERP system. Typically, each type of decision is associated with a specific level of the system and its controller. However, when the controlling software is written, much care has to be taken to ensure that the underlying and supervising controllers are suitably updated with the necessary information. • The current state of the system is captured via a manifold of sensors at various locations. The output of these sensors is then combined with knowledge about the system into a virtual representation of the systems’ status. • If adaptations of the systems are required, they have to be coordinated across all levels, requiring an intricate knowledge of the participating subsystems. This usually requires the involvement of people with different qualifications: software technology, electrical engineering, mechanical engineering, and possibly industrial engineering. Therefore, adaptations must be well prepared, thoroughly planned

24 Future Technologies in Intralogistics and Material Handling

551

Fig. 24.3 Overview of an example of a material handling system of manufacturing plant

and coordinated. In a business environment with frequently changing requirements and processes, material handling systems are not always suitably adaptable to such changing requirements. These deficits reduce the profitability of investments in automated material handling systems, and lead to more manual handling, since investments are tied up longer than the business environment is expected to remain stable. The challenge is to design more flexible material handling systems that can be easily adapted to new tasks or transferred to new locations. Research and industry push to fill this gap with newly developed technologies and material handling systems. These systems are representations of trends like Industry 4.0,1 Cyber-Physical Systems2 or Industrial Internet.3

1 German

coordinated initiative. of the National Science Foundation of the US. 3 International industrial initiative. 2 Initiative

552

K. Furmans et al.

24.1.2 Desirable Properties and Design Patterns of Future Material Handling Systems The lack of flexibility of today’s material handling systems has inspired several research institutes and material handling companies to develop a new kind of material handling systems. Furmans et al. (2010) have described the following desirable properties of future material handling systems: • “What You See is What You Get (WYSIWYG): The visible, physical system is all the user should care about. There should be no need to synchronize a physical layout with wiring and the software that contains the control logic. • Plug-and-Play (Plug-and-Work)-capability: Once the material handling system is physically configured, everything is done. New components are added by simple insertion. • Scalability: The system can be up- or downsized in order to adapt to changing performance requirements. • Reconfigurability: The operators must be able to change the system themselves, without the help of electricians or programmers. Changing the systems configuration must be possible within minutes, or at most a few hours. • Reliability through in-place replacement: In order to achieve high reliability, which is characteristic of tightly integrated material handling systems, failing components must be easily replaceable, and must not require resetting the whole system. The system detects its own failures and configures itself for repair. • Inherent safety: The system may not endanger the people around it. Transported or stored goods may never be damaged or lost. • Resource efficiency: Through easy reusability and by only operating those modules that are currently needed, less energy for operation and fewer resources for manufacturing the equipment are used. • Self-adaptability: The system should be able to adapt, within the limits of its physical representation, to changes in the patterns and quantities of the material flow. Ideally, the system should detect these changes in flow and be able to adapt itself accordingly. Furmans et al. (2010) have also deduced five design patterns that help to achieve the previously described desirable properties. • Modularity: The system should consist of highly independent modules, which supplement each other in order to perform the material handling task. The modules can be combined easily in order to create a system. The links between the modules are established by the modules themselves. • Function integration: Each module of the material handling system contains all functions necessary to perform its task. This usually includes, but is not limited to, identifying loads to be moved, deriving the destination in the system, recognizing the conditions of surrounding elements, moving goods in the appropriate direction, and passing appropriate information to surrounding modules or destinations.

24 Future Technologies in Intralogistics and Material Handling

553

• Decentralized control: The actions of the modules are controlled by their own controllers. • Interaction: Adjacent modules freely exchange information and goods. There is no central instance or master module with this as a special task. • Standardized physical and information interfaces: A major obstacle for reconfiguring automated material handling systems is the need to synchronize functionality of sensors, drives, controllers and mechanical components, which makes high-level interfaces to connect the modules with each other necessary. New systems should be able to exchange information on a function-based level, thus avoiding the problem of synchronizing parts of the system.” Seibold and Furmans (2016) have reviewed these desirable properties and design patterns and have found that the idea of Plug&Work-capable material handling systems has been widely spread during the last 5–10 years. In addition, the described properties and patterns fit very well into current technology trends for production systems (Industrie 4.0, Industrial Internet …). Seibold and Furmans (2016) have also presented several examples of connectionbased and trip-based material handling systems, which possess at least a few of the above presented capabilities of future material handling systems. They conclude that the development of connection-based material handling systems with Plug&Work-capability is closer to commercialization because trip-based material handling systems face additional challenges related to safety and navigation of free-moving vehicles. In the following section, we introduce functions that are required for material handling systems in general and example systems for future material handling. In the third section, we then discuss the development stage of each of these functions. We also investigate how the design patterns influence the required functions and to which degree the desirable properties are fulfilled already.

24.2 Required Functions and Examples for Flexible, Advanced Material Handling Systems (Advanced) Before introducing required functions for material handling systems, we first clarify differences between connection-based and trip-based material handling systems. Material handling systems can be categorized in connection-based systems, which are usually implemented as conveyors, and trip-based systems, which are using fixed or variable tracks. An Overhead Monorail is an example for a trip-based system with fixed tracks, AGVs or forklift-trucks are representatives of variable track systems. Usually, connection-based systems implement control logic, sensors and actuators in the conveyors (active and smart conveyor modules), while trip-based systems use smart vehicles and a simple infrastructure (rails, ground). Therefore, some of the functions are more complex to implement within trip-based material handling systems than with connection-based systems with a smart infrastructure. Table 24.1 lists

554

K. Furmans et al.

Table 24.1 Graph-based description of connection-based and trip-based systems Connection-based systems Trip-based systems Nodes/edges defined by

Conveyors

Not necessarily clearly defined (environment)

Mobile objects

Load carriers

Vehicles

Destination of mobile object

Specific for load carrier

Interchangeable

Active object

Nodes (conveyors)

Mobile objects (vehicles)

Table 24.2 Required functions of connection-based and trip-based material handling systems Planning Topology recognition and mapping Order (and energy) management path planning

Execution of transports Hardware control Load handover and identification Localization and navigation Reaction to objects in path Deadlock handling

General system functions Safety functions Human-machine-interaction Reconfiguration

the differences by a graph-based description. In connection-based systems, the nodes and edges of the system are defined by the physical arrangement of the conveyors whereas in trip-based system, they are not necessarily clearly defined if the vehicles are free-moving. In connection-based systems, load carriers are mobile objects that are moving through the defined graph and each mobile object has its specific destination. In trip-based systems, the destinations of the vehicles, i.e. the mobile objects, are interchangeable since they depend on the carried load. For developing control algorithms, it is relevant that in connection-based systems, the nodes of a graph are active objects whereas in trip-based systems, the mobile objects are active. The required functions for material handling systems are divided in three categories (see Table 24.2). The first category includes all functions that are necessary for planning a transport. The second category contains all functions for the execution of transports. In addition, the third category describes general system-wide functions. All functions are generally applicable to connection-based and trip-based material handling systems. For planning of a transport, the system already needs to know its topology and needs to create a map based on it. The map should include possible transport paths and information about the location of sources and destinations. The second function of planning is the order (and energy) management. This describes the decision which transport order should be executed by which transport resource and at which point of time. For the execution of transports, five functions are required: The hardware control is responsible for the control of sensors and actuators and is the lowest control level. The load handover and identification describes the function of handing over loads

24 Future Technologies in Intralogistics and Material Handling

555

from one material handling system to another one or to the environment (for example floor or human worker). Localization and navigation ensures that the transport resources or loads can be localized in the system/environment and that the moving objects follow the planned path. If there are obstacles, the moving objects need an appropriate reaction to objects in the path, which could be to wait or to replan the path in order to avoid collisions. And finally, deadlock handling is necessary if there are several moving objects of which the paths are crossing. In addition to the functions of planning and execution, there are three general system functions: First, the systems needs to have safety functions in order to guarantee human safety in its surrounding. Second, the system needs to have physical and informative human-machine-interaction. And third, a material handling system should have a function that allows for easy reconfiguration. All these functions are necessary in conventional just as well as in future material handling systems. But technological development enables to enhance the functions in order to increase flexibility of material handling systems and to achieve the desirable properties as described above.

24.2.1 Example Systems for Future Material Handling Over the last 5–10 years, several systems have been developed in research and industry following these design patterns. The following overview distinguishes connection-based and trip-based systems. The part on connection-based systems is based on Seibold (2016).

24.2.1.1

Connection-Based Systems

FlexConveyor is a modular material handling system with decentralized control (see Fig. 24.4). Each module is identical: a rectangular right-angle-transfer that is able to transport goods in the four cardinal directions, cf. Mayer (2009). Nowadays, different kinds of modules can be combined (Flexlog GmbH 2015). The conveying system can be easily built by the user by combining the conveying modules, because they are equipped with wheels and can be connected without any tools. The connection between neighbors is physical, electronic and electrical. Each module has its own control and communicates with its neighbors to take decisions. For each box entering the system, a route is reserved from source to destination to prevent opposing routes on bidirectional conveying modules. With so-called deadlock tokens, loops of boxes waiting for each other are avoided (Mayer and Furmans 2010). GridSorter is based on the idea of FlexConveyor. As shown in Fig. 24.5, the system is built of rectangular transfer modules (as in the early version of FlexConveyor), each being able to communicate with its neighbors and to take decisions (Seibold et al. 2013). The difference with FlexConveyor concerns the system task and the density of the conveying network. The main task of GridSorter is to sort goods, i.e.

556

K. Furmans et al.

Fig. 24.4 FlexConveyor, left: photograph of identical, transfer modules (Seibold et al. 2013), right: graphic of a network built of different module types (Flexlog GmbH 2015)

Fig. 24.5 Schematic representation of a GridSorter system

to transport goods to different destinations based on case-specific, known criteria. In order to reach a high module- and space-efficiency, the conveying network is built as densely as possible: the result is a grid-like network topology (Seibold et al. 2014). The topology may have any form and the sources and destinations can be at any side module of the conveying network. Seibold et al. (2013) adapt the routing and deadlock handling of FlexConveyor to GridSorter. With this control, deadlocks occur if high throughput is required and if order input is not controlled accordingly. GridStore and FlexConveyor are based on the same idea. In GridStore, a dense network of transfer modules with decentralized control, the grid, is used to store goods. GridStore aims to combine the apparently opposing objectives of high throughput and high density of a storage system (Gue et al. 2014). Items are retrieved on one side of the grid and replenished on the opposing side. The corresponding control algorithm is called “Virtual Aisles” because stored items are moved out of the way of requested items in order to form aisles. The decision as to how items should be moved is taken stepwise by each module based on the current situation. GridPick is a special implementation of GridStore. Items are retrieved and replenished on the same side of the grid in order to supply items of one order to a picker (see Fig. 24.6). In comparison to commonly used flow racks, the walking distance

24 Future Technologies in Intralogistics and Material Handling

557

Fig. 24.6 Graphic of GridPick (Uludag 2014) and GridSequence (Gue et al. 2012)

of a picker and therefore the pick time of one order can be reduced. To enable movements in all four cardinal directions, the “Virtual Aisles” algorithm has been adapted to balance the number of items per row (Uludag 2014). GridSequence again uses a dense network of transfer modules with decentralized control. The system task is to establish a certain sequence within the set of retrieved items (see Fig. 24.6). The “Virtual Aisles” algorithm has been adapted by assigning an intermediate destination to each item depending on its sequence number (Gue et al. 2012). Cognitive Conveyors consist of small-scaled modules with decentralized control (see Fig. 24.7). Since one module is as small as one roller, several modules must form groups in order to transport items: omnidirectional movement of items is possible. The main system task is to transport items, but it is also possible to use the system for buffering and sequencing (Krühn et al. 2013). Krühn (2014) describes the control algorithm which reserves a route from source to destination for each box and avoids deadlocks during transport.

24.2.1.2

Trip-Based Systems

KARIS PRO is a transportation system that consists of many AGVs with decentralized control. The vehicles coordinate themselves by negotiating with each other about which order is transported by which vehicle. Thus, a central control unit is obsolete (Furmans et al. 2014). The system is very flexible: By using natural landmarks for localization, KARIS PRO only needs a communication infrastructure and energy. This allows quick and easy implementation. One vehicle can move boxes, pallets or trolleys (see Fig. 24.8). Even though KARIS PRO is a trip-based system the vehicles can cluster to build a temporarily connection based conveyor system.

558

K. Furmans et al.

Fig. 24.7 Graphic of cognitive conveyors (Krühn 2014)

Fig. 24.8 Graphic of KARIS PRO

FiFi is a gesture controlled transportation robot (Trenkle et al. 2015). It uses data from a 3D-camera to detect people and gestures and supports different modes of operation (see Fig. 24.9): In Following Mode, FiFi follows a moving user by keeping a constant distance. Maneuvering Mode allows exact positioning of FiFi while following the movement of the users’ hands. In Cluster Mode several vehicles follow the user one after another to transport higher quantities of goods. For longer

24 Future Technologies in Intralogistics and Material Handling

559

Fig. 24.9 Graphic of FiFi

static transport routes without user-interaction, Line Following Mode allows FiFi to follow an optical track using a line-tracking camera.

24.3 Discussing Functional Challenges (State-of-the-Art) In this third section, we discuss all presented functions in detail and show challenges that are to be met in order to achieve the desirable properties. In the first part, we present five main functions that are needed for all advanced connection- or trip-based systems to make them work. In the second part, we describe remaining functions that are mainly needed for trip-based systems or are especially complex in trip-based systems.

24.3.1 Main Functions Needed in Connection-Based and Trip-Based Systems The required five main functions include mapping and topology recognition, needed to properly generate the graph describing the system. Within path planning, the route of the mobile objects is calculated which is then executed during transport with deadlock handling as an additional crucial function. Hardware control is needed to execute the movement of load carriers or vehicles and load handover and identification is needed at the beginning and the end of a transport.

560

24.3.1.1

K. Furmans et al.

Mapping and Topology Recognition

In today’s systems, the topology of a system i.e. the graph has to be modeled manually during installation of the system. This means that the position of sources and destinations and the routes between these points is predefined. Also, the position of other relevant points like charging stations is manually defined. This causes a high workload for modeling the system topology and keeping it up to date. In future material handling systems, the user should be able to change the setup of the system dynamically, which requires the system itself to recognize and model the current topology. In connection-based systems, the conveyor modules have be able to recognize their neighboring modules with the help of standardized interfaces. Sharing this information with all other modules, each module is able to deduce the network topology and to calculate shortest paths. Conventional algorithms like Link-StateRouting (Tanenbaum 2011) can be applied. In trip-based systems, topology recognition is much more complex, because it depends strongly on the kind of localization method. For high flexibility, AGVs should be able to freely moving around in their environment. Therefore, they should be able to create a map out of their sensor data. In a second step, this map needs to be interpreted: Nodes and edges have to be defined for path planning and special positions for load handover or battery charging need to be defined. KARIS PRO is able to update the map and topology based on the dynamically changing environment (Sun et al. 2016). To keep the map up-to-date and consistent, the vehicles exchange information about maps. Relevant desirable properties: WYSIWYG, Plug&Work, Reconfigurability Relevant design patterns: decentralized control, interaction, standardized information interfaces.

24.3.1.2

Path Planning

In todays’ material handling systems, path planning is usually done during installation manually. This means that a uni-directional network of fixed paths is defined, in connection-based as well as in trip-based systems. Therefore, during operation no decision on path planning is necessary any more if a load or a vehicle needs to move from A to B. In Fig. 24.10, this corresponds to the left category of routing: the forwarding of loads or vehicles on a defined path. Considering modular connection-based systems, Seibold (2016) has presented three other routing strategies promising more flexibility by planning a path for each individual object enabling adaptive behavior to changing requirements. For example, a loop of conveyors can be used bi-directionally with direct routes when utilization is low whereas it is used one way when traffic increases. Within dynamic routing, the routing decision is step-wise taken during the transportation of a load. If the path is planned before the start of a transport, time-

24 Future Technologies in Intralogistics and Material Handling

561

Routing

for each individual object

for a group of objects

dynamic

Forwarding

Dynamic Routing

static

time-independent

time-window-based

Time-independent Route Reservation

Time-window-based Route Reservation

Fig. 24.10 Classification of routing strategies (Seibold 2016)

independent or time-window-based route reservation can be used. By reserving the planned path, the interference of different loads can be considered. In trip-based systems with decentralized control, path planning can be done in a decoupled way which means that each vehicle plans its path assuming that there is no other vehicle interfering (Sun et al. 2014). In both kinds of systems, traditional path planning algorithms like those shown in Fig. 24.11 can be applied. For adapting these algorithms in connection-based systems, it is important to note that the active objects i.e. the nodes of the graph do the path planning. Seibold (2016) has adapted Depth-First Iterative Deepening for time-window-based reservation to decentralized control resulting in the so-called DIDA*-algorithm. Relevant desirable properties: Plug&Work, Reconfigurability, Scalability, Selfadaptability Relevant design patterns: decentralized control, interaction.

24.3.1.3

Deadlock Handling

Deadlock handling is necessary if moving objects, i.e. loads or vehicles might block each other in such a way that no object is able to move anymore. Figure 24.12 shows two example situations of deadlocks in connection-based systems. These situations are also possible in trip-based systems. Seibold (2016) describes four different deadlock handling strategies (compare to Fig. 24.13) based on the categorization of Tanenbaum and Bos (2015) for operating systems. Ignoring deadlocks is only possible if it is not very likely that deadlocks happen, for example in a one way network without loops that are likely to be filled up with

562

K. Furmans et al.

Known Path Costs

Breadth-First

Dijkstra

Predictive Heuristic

Best-First Search with

Uninformed Search

Tree of Partial Paths

A*

One Partial Path

Depth-First Iterative Deepening

Depth-First

IDA*

Fig. 24.11 Basic categorization of existing path planning algorithms (Seibold 2016)

Fig. 24.12 Deadlocks in connection-based systems (Seibold 2016)

transported loads. In conventional connection-based and trip-based systems, this is usually the case: the systems are designed in such a way, that all edges are unidirectional and loops—if existing—are large enough to avoid being filled completely. This limitation clearly contradicts the desirable properties of WYSIWYG and Plug&Work: The user should be able to design the material handling system freely without being forced to consider design limitations and without the need to define fixed moving directions in the complete system. For this reason, new deadlock handling strategies have been developed for connection-based material handling systems with decentralized control. In some systems, deadlocks are allowed to occur but are subsequently detected by the modules which have strategies to recover, for example by backtracking. Avoiding deadlocks requires intensive communication and complex decision making during transport. For further information, we refer to Mayer (2009) or Krühn (2014).

24 Future Technologies in Intralogistics and Material Handling

563

GridStore

Prevent GridSorter

GridSequence

Avoid FlexConveyor

Cognitive Conveyor

GridPick

GridFlow

Detect & Recover

Cellular and Modular Warehouse Internet of Things

EHB

Ignore

low goods density sparse & unidirectional network

sparse & bidirectional network

high goods density dense & bidirectional network

Fig. 24.13 Classification of deadlock handling strategies in connection-based material handling systems with decentralized control (Seibold 2016)

Seibold (2016) has transferred the principle of Logical Time (Lamport 1978) from distributed systems to material handling systems. Each conveyor module owns its own logical clock and locally assigns timestamps to transport steps, which are then used to establish a partial order among all transport steps in the system. The logical clocks of neighboring modules are synchronized if a transport takes place. Like this, deadlocks are prevented i.e. structurally negated, cf. Seibold (2016). As already described, in free-moving trip-based systems, each vehicle usually executes path planning in a decoupled way. This means that local priority rules are needed in order to define the behavior of vehicles of which the routes interfere with each other as will be discussed in the function Reaction to Objects in the Path. To resolve conflicts, waiting positions are necessary where vehicles do not block other vehicles. The handling of loop deadlocks with multiple vehicles is not a main focus in trip-based systems because, usually, the number of vehicles is limited in order to guarantee efficient system behavior. The number of vehicles is also small compared to the number of potential vehicle locations, resulting in a low probability (if at all) of deadlocks. Relevant desirable properties: WYSIWYG, Plug&Work, Self-adaptability Relevant design patterns: decentralized control, interaction.

564

K. Furmans et al.

Layout Manager

Reservation Manager

Transport Manager

Path Planning

Deadlock Handling

Topology Recognition

Hardware Control

Performing transports based on sensor data and by controlling actuators

Logical Clock

Fig. 24.14 Control components of GridSorter (Adapted from Seibold 2016)

24.3.1.4

Hardware Control

The hardware control is responsible for the control of sensors and actuators and is the lowest control level. Therefore, its main responsibility is to control the movement of loads or vehicles while sensing its environment. In today’s systems, the system control is strictly hardware-dependent: It is not possible to exchange single hardware components, for example a motor, or to change the transport tasks of a module without a significant effort to adapt the complete system control as described in the first section. In future material handling systems, we also apply the principle of Plug&Workcapability to the hardware and control components of one module or vehicle. Therefore, we need the same design patterns: modularity, function integration and standardized physical and information interfaces. This means that the hardware and control components in one module should be (as much as possible) independent. Most of the control components are naturally hardware-independent by abstracting functions like path planning (compare Fig. 24.14). The information interfaces between control components should be on a high logical level to enable easy integration. In connection-based systems, FlexConveyor is a good example of how to achieve hardware-independent control by looking at a system with different kind of conveyorbased modules like curves, linear modules or crossings (as right-angle transfers) with a varying number of in- and outputs. The hardware control is designed in such a way that it only needs to be parameterized once at the end of production when the shape of the module, the position of motors and sensors and the resulting in- and outputs are known. This allows the user to design individual conveying modules without the need to program a specific hardware control for each module type. For trip-based systems, KARIS PRO is an example that demonstrates hardwareindependent control. Most of the control components are hardware-independent.

24 Future Technologies in Intralogistics and Material Handling

565

The motors are controlled by an independent control component that calculates the required velocity of each motor based on the movement vector defined by the navigation. As long as the hardware control knows the abilities of the drive steering unit and is able to communicate those to the navigation, it is possible to change the drive-steering unit (for example from omni-wheels to normal wheels) without the need to adapt the remainder of control components. Relevant desirable properties: Plug-and-Play-capability Relevant design patterns: modularity, function integration, standardized physical and information interfaces.

24.3.1.5

Load Handover and Identification

Load handover and the identification of the load is needed if two material handling devices interact with each other or if a load needs to be handed over to the environment. Generally, there exist different cases: • The load is handed over from one active material handling device to another active device. • There is only one active device handing over the load to a passive device, for example a storage location. • A human worker is responsible for the load handover. If both devices are active, an exact synchronization of different controls (cf. Sect. 24.1) is required, which results in a high installation effort in today’s systems. In addition, the load handling devices in today’s systems are usually specialized for a certain kind of load carrier, i.e. for a special range of size, weight and shape of boxes or products. To achieve the desirable properties Plug&Work-capability and self-adaptability, the function of load handover is crucial: The installation or relocation of passive or active load handling devices should allow for easy or even self-adaptation to changing requirements. To achieve complete function integration, each module should be able to identify the load. In order to save costs, as an alternative, identity information can also be handed over from one module to another parallel to the physical material flow. For high flexibility, the material handling system should also be able to handle different kind of loads, for example boxes and pallets. In connection-based systems, standardized physical and information interfaces and function integration enable a load handover: If every module is able to fulfill a transport activity with its neighboring module, load handover is an integrated system function. If the modules are arranged in a dense surface as for example in GridSorter, GridStore or Cognitive Conveyors, even different sizes of loads can be handled. Here the challenge lies in the joint path planning (see Krühn 2014) as well as in the control of the involved actuators. In trip-based systems, load handover is a bigger challenge: To reduce installation and synchronization effort, the load handover should be performed with one passive

566

K. Furmans et al.

device. This device should be easily installable or should be part of the general environment, for example a storage location on the floor or in a rack. On the other hand, this often increases the duration of load handover and the complexity of the load handling device. In order to handle different kinds of load, modularity is again a helpful design pattern: if the vehicle is able to change its load handling device and even cooperate with other vehicles as in KARIS PRO, the system is able to adapt itself to changing requirements. Relevant desirable properties: Plug-and-Play-capability, self-adaptability Relevant design patterns: Interaction, standardized physical and information interfaces, modularity.

24.3.2 Functions with Special Complexity in Trip-Based Material Handling Systems The following six functions represent the complexity existing in trip-based systems. Order management is necessary because transport orders of loads need to be assigned to vehicles in trip-based systems. Safety functions and human-machine-interaction are crucial if vehicles move in the same areas as human workers. Localization and navigation and the reaction to objects in the path are only necessary in trip-based systems because the edges of the graph are not as clearly defined and exclusive as in connection-based systems. And finally, reconfiguration is mainly possible in trip-based systems because they are able to rearrange by movement of vehicles.

24.3.2.1

Order and Energy Management

Order management refers to the allocation of resources to transport orders which includes the decision which resources are required at which time to execute the order. For connection-based systems, the path planning mechanisms provide this functionality. Energy management is not crucial since connection-based systems usually have a permanent connection to the power grid. Power saving strategies can be implemented, based on the path planning results, indicating when a conveyor should be switched off and on. The decision parameters regarding how much energy is consumed in idle time and when a power down and a power up cycle would be more efficient can be determined once and then implemented in the local control. Trip-based systems offer more flexibility and therefore a higher complexity in order and energy management. A vehicle should be made available for order management by a clear status (busy, blocked, disabled) to all other vehicles. It should be easy to add a vehicle to the pool of available resources or to remove one vehicle. This requires wireless technologies like WiFi, Bluetooth, or NFC and an easy but secure authorization mechanism for the management of the pool of vehicles. This also pro-

24 Future Technologies in Intralogistics and Material Handling

567

vides the capability to scale the system capacity up and down as well as to reconfigure it in the sense of managing different types of vehicles. The same is true for “in place replacement”. Today, order management for trip-based systems is done by a central system control, which assigns orders to vehicles. Several algorithms are used for the assignment of orders to vehicles, from a simple FIFO-assignment of orders to the next free vehicle to the continuous, repeated solution of a vehicle routing problem. The next level of decentralization can be reached by treating each vehicle as an agent that performs the planning and coordinates with other vehicles by announcing its plans and negotiation options. The current focus in research is on decentralized planning with implicit coordination. Coordination is achieved through the observation of the other vehicles, assuming that if the individual plans have certain properties, then efficient plans can be implemented without explicit coordination mechanisms. These algorithms are partly inspired by biological systems, like transport through ants or nectar collection by bees. Closely linked to the order management task is the dwell point strategy: where should vehicles wait, if there is no open transport order. If there is a central controller, this controller assigns the dwell point to the vehicle. Energy consumption in the short term can be minimized, by letting the vehicles wait at the end point of their last transport order, thus wasting no energy on trips, which might not be needed. Another option is the long-term observation—or even planning—of the transport orders. From this, a matrix of transports can be derived. By comparing the number of arrivals at a destination with the number of departures, we may determine whether a node in the network rather needs empty vehicles or provides empty vehicles. By solving the transport problem, using information about the trip-costs between nodes, we can compute the most cost-efficient destination for empty vehicles. This however assumes that timing issues are of minor importance. However, if it is required that the material handling systems can react quickly and start the next transport order, it is useful to move the vehicle to a position where the next transport order is most likely to come up. A stochastic optimization problem must be continuously solved, where the likelihood of a new order over time at a location is compared to the time that a free vehicle would need to reach this area. From the minimization of the expected waiting time of the transport orders, the trips of the empty vehicles can then be deduced. Resource efficiency can be influenced by using and operating only the number of vehicles that are required and relocating vehicles to other operations or returning them to a leasing and renting company. Relevant desirable properties: Plug&Work, self-adaptability Relevant design patterns: decentralized control, interaction.

568

24.3.2.2

K. Furmans et al.

Safety Functions

Safety functions are required to protect humans at the operating area of material handling systems. There are various hazards e.g. impact from dropped load or collisions, crush from colliding totes or drawing-in from conveyor belts. In today’s connection-based systems, mechanical protective measures like guide rails are used to prevent totes from falling down or finger protections to prevent crushing fingers between conveyor belts. Central emergency stopping devices stop the system as soon as an emergency button is pressed. However, every time the system structure is changed, the safety system also has to be adapted. In today’s trip-based systems, sensors like bumpers for slowly operating vehicles with rather short stopping distance or expensive safety laser scanners with protective fields for fast operating vehicles are used to prevent collisions. The protective fields trigger the safety brake if any obstacle reaches the area. Alternatively, instead of protective systems at the vehicles, the operating areas of the vehicles are physically separated from human working spaces. In this case, if a worker needs to enter the system’s operating area, the whole system has to be stopped. This results in high costs for downtimes. Another measure concerns the securing of loads, which is necessary to prevent the load from falling down while braking or driving along curves. To achieve Plug&Work-capability, the integration of safety functions into the modules is essential. Instead of using one centralized safety system, the safety related sensors, controllers and actuators are directly integrated within the affected components. Integrated safety reduces wiring and guarantees reconfigurability of the system without manual adaptation of safety functions. In connection-based systems, safety integrated in modules results in more flexibility. The user can rearrange the conveyor modules according to his needs without adapting or reinstalling the safety components. In trip-based systems, single modules like the drive unit can easily be exchanged or upgraded without making any changes to a central safety system. Modularity therefore requires a deep integration of safety, which often enables the safety components to acquire redundant data from attached components. Safety measures or decisions handling unsafe situations are then to be made within the modules. The modular concepts of safety and hardware control depend on each other. To design modular systems one should regard the possible applications, re-arrangements of modules and the resulting hazards while performing a risk-assessment. As Plug&Work-capable trip-based systems make autonomous decisions about their paths, they also require more complex safety functions. Additionally to safety functions that prevent collisions safety functions need to protect the vehicles from falling down ramps or stairs (Trenkle et al. 2013). Relevant desirable properties: inherent safety Relevant design patterns: modularity, function integration.

24 Future Technologies in Intralogistics and Material Handling

24.3.2.3

569

Human-Machine-Interaction (Physical and Information)

Today’s material handling systems mostly use pragmatically designed, simple user interfaces like buttons and light signals. Touch panels are used in control cabinets for visualization of the system and for diagnostics purposes. Developers build the interfaces with the following question in mind: Which information is required for the process to work? We instead propose asking: How can machines and natural persons understand each other? Therefore, we need to understand the characteristics of a good user interface. Leonhard (1999) calls simplicity and self-expression the most important characteristics of a user interface. This includes intuitive rules for the user, which make special training or the reading of a manual superfluous. Nielsen’s (1994) usability heuristics complement the minimal burden of memory, the consistency of input and output, and the return of feedback. Raskin (2000) describes two laws of interface design in accordance with the Asimovian robot law. The first relates to the preservation of work and reads: “A computer shall not harm your work or, through inactivity, allow your work to come to harm”. Raskin’s second law of interface design concerns effectiveness: “A computer shall not waste your time or require you to do more work than is strictly necessary”. How can we apply this to Plug&Work-material handling systems? Regarding Golden Krishnas statement “The best interface is no interface” (Krishna 2015), we suggest to implement user interfaces with care. One should well consider if an explicit interface is necessary or if implicit communication is available to make decisions without explicit user input. In both connection-based and trip-based systems, it is necessary that the user can predict and understand the systems’ actions and on the other hand, the system should be able to understand human actions (Trenkle et al. 2017). Following the WYSIWYG principle, the user is able to interact with Plug&Work connection-based systems by re-arranging conveyor modules and putting goods on the system. Visualization of system topology and location of load carriers can be done by a mobile web interface that collects all information of the decentralized controls of conveyor modules. In trip-based systems, interaction affects not only the users, who purposely interact with the systems, but also other workers who happen to be on the driveways of the vehicles. Humans, encountering the system should be able to foresee the next actions of a vehicle. This includes driving maneuvers or operations like load handling. Therefore, interfaces like optical signals on the vehicles are required. When designing the signals one should take care they can be understood without reading the manual. This can be reached by using signals from daily experiences for example using blinker signals or red/white driving direction lights that are well known in daily traffic. Using gestures or speech from human-human interaction is key for building natural user interfaces. FiFi, for example, uses a wave gesture to make the vehicle following the user. Future trip-based systems do not need explicit user input: by putting a load carrier on a vehicle, the vehicle reads the RFID-tag and derives its destination from the material. It can also detect whether the user is close and only starts

570

K. Furmans et al.

the transport if the path is free. Thus, implicit understanding provides interaction without dedicated user interfaces. For the visualization of system status information beyond that of a single vehicle, we suggest using mobile interfaces. They should provide information about the system state, its vehicles or conveyors and performance data. Promising hardware solutions are tablet computers or head-up displays for virtual or augmented reality. Wearables can be used to send critical system states to the user. Relevant desirable properties: WYSIWYG, Plug&Work Relevant design patterns: interaction.

24.3.2.4

Localization and Navigation

Conventional trip-based transportation systems usually follow fixed tracks. They use lines, magnets and barcodes on the floor or reflectors on the wall to determine their position. These methods are robust but have several drawbacks like high installation costs due to changes on the environment. Any changes in routes require adaptations to the markers. Especially in systems with many vehicles, a central unit coordinates the navigation. For Plug&Work systems localization has to cope with dynamic environments without using artificial landmarks. This means, that localization and navigation has to work even if a high percentage of the natural landmarks have changed. Sun et al. (2016) present a dynamic localization approach based on SLAM that allows a robust localization in different real-world environments by only using laser-scanners and odometric data. For navigation, the vehicle calculates a resulting movement vector based on the result of localization and the previously planned path. This movement vector is continuously fed to the hardware control, which then executes the desired movement. In case of occurring obstacles, the navigation needs to react to the objects described in the following section. Relevant desirable properties: Plug&Work, Self-adaptability Relevant design patterns: Function Integration, Interaction.

24.3.2.5

Reaction to Objects in Path

Conventional trip-based systems provide simple strategies to handle obstacles. Sensors indicate the vehicle to stop until the obstacle disappears. Unfortunately, most trip-based systems cannot detect obstacles under or over the detection range of the laser scanners. This can also result in collisions and damage of vehicle or load. Therefore, trip-based Plug&Work material handling systems can handle obstacles more effectively. After detecting an obstacle in the planned path, a decision is made based on the available space and information from the map. This includes replanning the path to navigate around the obstacle. By using 3D-sensors in the future, a vehicle

24 Future Technologies in Intralogistics and Material Handling

571

can decide if it can pass under an obstacle considering the size of the load or if a new path has to be calculated. Plug&Work material handling systems not only differentiate between free driveways and obstacles. They also classify the obstacles and consider the surroundings to provide reactions that are more adequate. Large robot-teams communicate and use predefined traffic rules to avoid collisions and deadlocks (Sun et al. 2016). Special rules depending on the specific situation enable efficient system behavior in intralogistics. For example, fast moving obstacles on driveways can be ignored if the distance increases. While sometimes overtaking other vehicles is useful, it should be avoided in areas like loading zones for milk runs if it interrupts the runners’ loading process. When driving on workers’ travel paths, the vehicle should show foreseeable behavior and communicate its actions. If driving towards a walking person, the vehicle should always attempt to plan a path on the right side and use visual signals to signal an evasive maneuver. • Relevant desirable properties: Plug&Work, Reliability • Relevant design patterns: Interaction.

24.3.2.6

Reconfiguration (Software and Hardware)

Reconfiguration of material handling systems can be divided in two categories. On the one hand, control algorithms should adapt to changing requirements like for instance more or less throughput for certain sources or destinations. On the other hand, system topology should adapt to changing requirements like repositioning of sources and destinations. In today’s material handling systems, neither control nor topology reconfiguration is possible without a large installation effort. If future material handling systems follow the described design patterns, manual reconfiguration of topology becomes possible achieving flexibility in automated systems. Nevertheless, this reconfiguration is still manually driven which means that the user first needs to detect the need of reconfiguration, followed by its execution. Reconfiguration of control currently partly exists as path planning for each individual load and reaction to objects in paths but all behavior rules are pre-defined and so far are not self-learning or adaptive. Future research and development should focus on the design of control algorithms that do not just focus on functioning system behavior but also support the optimization of system behavior. This means that all modules or vehicles should analyze the resulting system performance in order to refine existing control algorithms. In connection-based systems, reconfiguration of topology will always remain manual. However, the need of reconfiguration should be detected by the system itself. In trip-based systems, the possibility of topology recognition strongly depends on the functionality of load handover and the load handling device. If the vehicles are able to handover load directly to the environment, the system is able to adapt itself to changing requirements. In addition, the functionality of cooperation of several

572

K. Furmans et al.

vehicles in order to transport big loads or to form temporal conveyors enables the system to high adaptation. • Relevant desirable properties: Self-adaptability • Relevant design patterns: Modularity, Function Integration.

24.4 Further Reading Developments in automation for production and logistics system have been increasingly dynamic in recent years. Especially in the field of AGVs, many different companies have responded to the increasing demand of flexible material flow in the future smart factory. The development of flexible automated systems has received quite some attention in recent years (Gilchrist 2016; Jeschke et al. 2016). In addition, we have been involved in the development of several connection-based and trip-based systems, based on these design patterns, in which the increasing complexity of each single function raised challenging design questions. This chapter serves to emphasize the urgent need to pay attention to a well-thought MHS design while providing a glimpse on the complexities of associated control and hardware design. For more detailed information on these requirements, and on control algorithms in particular, the reader is referred to: Material Handling and Logistics Roadmap 2.0: www.mhlroadmap.org Mapping, Localization and Navigation: (Parker 2009; Thrun et al. 2005) Order Management: (Le-Anh and De Koster 2006; Qiu et al. 2002; Weyns et al. 2008) Hardware Control: (Stichweh 2017) Sensors and Identification: (Fürstenberg and Kirsch 2017) Deadlock Handling: (Krühn 2014; Lamport 1978; Mayer and Furmans 2010; Tannenbaum and Bos 2015) Safety Functions (Börcsök 2007; Trenkle et al. 2013) Human-Machine-Interaction: (Raskin 2000) Automated Guided Vehicles: (Ullrich 2014) In our opinion, digitalization and automation of factories needs further innovative and extensive research and development. If material handling systems should be able to efficiently perform any transport task, they need to be more adaptive to different types of loads and load handover and changing requirements. In short, material handling systems should become more human-like: better interaction with humans and machines, self-sensing and self-adaptation to any requirement.

24 Future Technologies in Intralogistics and Material Handling

573

References Arnold D, Furmans K (2009) Materialfluss in logistiksystemen. Springer Börcsök J (2007) Functional safety: basic principles of safety-related systems. Hüthig Flexlog GmbH (2015) Dezentral steuerbar—Modulbaukasten mit FlexTechnology. www.flexlog.d e/de/produkte/flextechnology. Accessed 23 July 2015 Furmans K, Schönung F, Gue KR (2010) Plug-and-work material handling systems. In: Proceedings of the international material handling research colloquium Furmans K, Seibold Z, Trenkle A, Stoll T (2014) Future requirements for small-scaled autonomous transportation systems. In: Production environments proceedings of the 7th international scientific symposium on logistics, Logistics Journal Fürstenberg K, Kirsch C (2017) Intelligente Sensorik als Grundbaustein für cyber-physische Systeme in der Logistik. In Handbuch Industrie 4.0 Bd. 3. Springer, Berlin, Heidelberg, pp 271–297 Gilchrist A (2016) Designing industrial internet systems. In Industry 4.0. Apress, pp 87–118 Gue KR, Furmans K, Seibold Z, Uludag O (2014) Grid-Store: a puzzle-based storage system with decentralized control Gue KR, Uludag O, Furmans K (2012) A high-density system for carton sequencing. In: Bundesvereinigung für Logistik (eds), proceedings of the 6th international scientific symposium on logistics (ISSL) Jeschke S, Brecher C, Song H, Rawat D (eds) (2016) Industrial internet of things: cyber manufacturing systems. Springer Krishna G (2015) The best interface is no interface: the simple path to brilliant technology. Pearson Education Krühn T, Radosavac M, Shchekutin N, Overmeyer L (2013) Decentralized and dynamic routing for a cognitive conveyor. In: IEEE/ASME proceedings of the international conference on advanced intelligent mechatronics (AIM), pp 436–441 Krühn T (2014) Dezentrale, verteilte Steuerung flächiger Fördersysteme für den innerbetrieblichen Materialfluss. Dissertation, Gottfried Wilhelm Leibniz Universität Hannover, Hannover Lamport L (1978) Time, clocks, and the ordering of events in a distributed system. Commun ACM 21(7):558–565 Leonhard JF (ed) (1999) Medienwissenschaft: ein Handbuch zur Entwicklung der Medien und Kommunikationsformen (vol 15, no 1). Walter de Gruyter Le-Anh T, De Koster MBM (2006) A review of design and control of automated guided vehicle systems. Eur J Oper Res 171(1):1–23 Mayer S (2009) Development of a completely decentralized control system for modular continuous conveyor systems. Dissertation. Universität Karlsruhe Mayer S, Furmans K (2010) Deadlock prevention in a completely decentralized controlled materials flow systems. Logist Res 2(3–4):147–158 Nielsen J (1994) Usability engineering. Elsevier Parker LE (2009) Path planning and motion coordination in multiple mobile robot teams. Encyclopedia of complexity and system science, pp 5783–5800 Qiu L, Hsu WJ, Huang SY, Wang H (2002) Scheduling and routing algorithms for AGVs: a survey. Int J Prod Res 40(3):745–760 Raskin J (2000) The humane interface: new directions for designing interactive systems. AddisonWesley Professional Seibold Z, Stoll T, Furmans K (2013) Layout-optimized sorting of goods with decentralized controlled conveying modules. In: Proceedings of the 7th annual systems conference (SysCon). IEEE, pp 628–633 Seibold Z, Gebhardt M, Stoll T (2014) Mehr Nutzen mit dem GridSorter. Hebezeuge Fördermittel 2014–5:260–262 Seibold Z (2016) Logical time in decentralized control of material handling systems. PhD thesis. KIT Scientific Publishing, Karlsruhe. ISBN 978-3-7255-0567-9

574

K. Furmans et al.

Seibold Z, Furmans K (2016) Plug&Play-Fördertechnik in der Industrie 4.0. In: Vogel-Heuser B, Bauernhansl T, ten Hompel M (eds) Handbuch Industrie 4.0, pp 1–17. Springer, Berlin, Heidelberg Stichweh H (2017) Aktorik für Industrie 4.0: Intelligente Antriebs-und Automatisierungslösungen für die energieeffiziente Intralogistik. In Handbuch Industrie 4.0 Bd. 3, pp 249–269. Springer, Berlin, Heidelberg Sun D, Kleiner A, Nebel B (2014) Behavior-based multi-robot collision avoidance. In: 2014 IEEE international conference on robotics and automation (ICRA), pp 1668–1673. IEEE Sun D, Geißer F, Nebel B (2016) Towards effective localization in dynamic environments. In: 2016 IEEE/RSJ international conference on intelligent robots and systems (IROS), pp 4517–4523. IEEE Tanenbaum AS (2011) Computer networks, 5th edn. Pearson Education, Boston Tanenbaum AS, Bos H (2015) Modern operating systems, 4th edn, (Global edition ed.). Pearson, Boston, Massachusetts Thrun S, Burgard W, Fox D (2005) Probabilistic robotics. MIT Press Trenkle A, Seibold Z, Stoll T (2013) Safety requirements and safety functions for decentralized controlled autonomous systems. In: 2013 XXIV international symposium on information, communication and automation technologies (ICAT), pp 1–6. IEEE Trenkle A, Göhl M, Furmans K (2015) Interpretation of pointing gestures for the gesture controlled transportation robot “FiFi”. In: 2015 9th annual IEEE international systems conference (SysCon), pp 721–726. IEEE Trenkle A, Furmans K (2017) Der Mensch als Teil von Industrie 4.0: Interaktionsmechanismen bei autonomen Materialflusssystemen. In Handbuch Industrie 4.0 Bd. 3, pp 45–59. Springer, Berlin, Heidelberg Ullrich G (2014) Automated guided vehicle systems: a primer with practical applications. Springer Uludag O (2014) GridPick: a high density puzzle based order picking system with decentralized control. Dissertation, Auburn University, Auburn, Alabama Weyns D, Holvoet T, Schelfthout K, Wielemans J (2008) Decentralized control of automatic guided vehicles: applying multi-agent systems in practice. In: Companion to the 23rd ACM SIGPLAN conference on object-oriented programming systems languages and applications, pp 663–674. ACM

Chapter 25

Supply Chain Security Gerwin Zomer

Abstract In this chapter, we introduce supply chain security and describe how governments as well as the logistics industries, shippers and manufacturing industries attempt to safeguard supply chains against theft, pilferage, terrorist attacks or unforeseen events that harm the seamless, reliable and efficient flow of goods. We define the concept of supply chain security and describe the context (basic part). Next, we elaborate how organizations manage corresponding supply chain risks by using models, methodologies and common practices (basic). Then, we discuss the government perspective and describe how supply chain security is embedded in laws and regulations and how business and government collaborate in joint supply chain security programs (advanced). In addition, we describe innovative control and supervision models that maximize the potential of collaboration with trustworthy supply chain partners that are in control of activities across their supply chains (state-of-the-art). This chapter ends with an elaboration of the digitization of global trade and concomitant cyber security threats (state-of-the-art).

25.1 Understanding Supply Chain Security (Basic) International trade is a key factor of global economic growth. The globalized economy is characterized by fragmented value chains involving many actors and complex supply chains. These supply chains are exposed to sophisticated criminal activities, such as theft, counterfeiting, smuggling, and terrorist support activities. Society and the global trade system must be protected against such activities. Therefore, supply chain security is a societal challenge. However, it is also a business challenge. Protecting the integrity of end-to-end supply chains has led to complex cross-border movements (Thomas 2010). Companies are highly dependent on their supply chains. With supply chains getting more fragmented, the possibility of supply chain disruption increases. Besides, when disG. Zomer (B) TNO, Delft, The Netherlands e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_25

575

576

G. Zomer

ruptions arise, companies need to have successful measures to prevent the loss of market share. Therefore, supply chain security is often integrated in the supply chain risk management framework of individual companies. Because a chain is only as strong as its weakest link, this is also a relevant statement for supply chain security. Corporate risk management frameworks focus on the risk exposure for the enterprise, using internal control frameworks with a limited endto-end scope. Supply chain security is often embedded in standard supply chain management practices. It includes activities such as screening the credentials of supply chain actors and the cargo, ensuring the integrity of the cargo in transit and in rest, and advanced notifications to support effective inspection processes on entry or at destination (Gupta 2015). Integrity in transit is being assured through physical or digital locks and seals. When cargo is in rest, integrity is safeguarded through gates, fences and other entry and access restrictions. It requires a collaborative approach to achieve end-to-end security. The above observations highlight the complexity of managing supply chain security in an effective and efficient way. Supply chain security is driven by a combination of societal interests, corporate interests and interests of the value chain of cargo being transported. Authorities and governments have recognized this and in addition to a regulatory framework with restrictive policies and procedures for supply chain security, a number of collaborative initiatives and programs such as C-TPAT, AEO and TAPA have been launched in the last few decades to jointly address the supply chain security challenges together with the trade and logistics industry (see also Chap. 7).

25.2 The Business Perspective of Supply Chain Security (Basic) For companies, supply chain security is managed in two ways, as an integrated part of supply chain risk management and of trade compliance management. Supply chain risk management is driven by commercial interests, whereas compliance management is driven by requirements set by governments or authorities. The latter will be discussed in a later section (see also Chap. 7 which is entirely devoted to compliance issues). This section deals with supply chain security as part of supply chain risk management. Supply chain risk management aims to minimize the possible occurrence of incidents or fraudulent acts regarding inbound supply that may lead to a financial loss for the (purchasing) firm (Zsidisin 2005). Talbot and Jakeman (2009) defines security risk as “any event that might result in the compromise of organizational assets i.e., the unauthorized use, loss, damage, disclosure or modification of organizational assets for the profit, personal interest or political interests of individuals, groups or other entities”. Supply chain risk management therefore includes the following components:

25 Supply Chain Security

• • • • •

577

Risk identification and modelling Risk analysis and impact assessment Risk management Risk monitoring and evaluation Organizational and personal learning including knowledge transfer.

It starts with integrating these components in your supply chain design, deciding where to source, consolidate and deconsolidate and where to keep inventories. To understand these components and their application in supply chains, we need to understand the difference between threats, vulnerabilities and risks. A threat can exploit a vulnerability, intentionally or accidentally, and obtain, damage, or destroy an asset. Threats need to be identified, but they often remain beyond our control. Examples include a hurricane or a tsunami, they cannot be prevented or tamed in advance. Vulnerability can be treated by identifying weaknesses and proactive measures. Risks can be mitigated by either lowering the vulnerability or the overall impact on the business. So what are the most apparent threats and vulnerabilities in supply chain risk management? We can distinguish two type of supply chain risks: risks arising from interactions within the supply chain and risks arising from interaction between the supply chain and its environment. The first type includes changes in customer demands, unexpected transit delays, production problems at suppliers of critical components and warehouse shortages. They are caused by lack of visibility, lack of ‘ownership’, justin-time practices and inaccurate forecasts. The second type includes external risks, such as disruptions caused by strikes, terrorism and natural catastrophes (Cranfield 2002). Together, they impact the vulnerability of the supply chain. Key concerns of supply chain managers are related to product quality, inventory and natural disasters. It takes years to build a reputation of delivering high quality products and services; recovery from quality problems is hard. Global production and sourcing has created longer supply chains putting pressure on safeguarding fast deliveries whilst putting pressure on minimizing inventories for optimal working capital and cash flow. This makes the supply chain vulnerable for disruptions with immediate impact on a companies’ promise to deliver capabilities. A third concern is the disturbing effect on supply chains of natural disasters such as the Iceland volcano disruption, the Japanese tsunami, and the flooding in Bangladesh or Thailand. These external factors can have a devastating impact on the efficiency and reliability of fulfilment processes. Terrorism and piracy, and delay caused by customs are less concerning for supply chain professionals. According to Dittmann (2014), compliance and security programs have resulted in faster customs processing. Recent studies highlight the growing concern of cyber security for supply chain management (Voster and De Bruijn 2016). This topic is being discussed later. Companies use risk analysis methods to identify and assess risks and determine effective strategies to cope with the risks in a structured way. Failure mode and effect analysis (FMEA), is such a method. Three factors define the priority: the likelihood or

578

G. Zomer

Fig. 25.1 Example of a bow-tie diagram

probability of the problem occurring, the seriousness or severity of the consequences, and the likelihood of early detection of the problem. Fault Tree Analysis (FTA) is a proactive method to identify all possible causes of a potential incident. The method helps to understand how systems can fail and to identify ways to improve system reliability. Fault trees have strong similarity with methods used in quality management, such as fishbone or Ishikawa diagrams. Where FTA identifies the causes, Event Tree Analysis (ETA) analyses the consequences of a potential event. The Bow Tie diagram is a combination of both fault tree and event tree. An example of a bow-tie diagram is presented in Fig. 25.1. Given a specific hazard, prevention measures are placed in the fault tree and mitigation measures in the event tree. The Risk Exposure Index is a method developed by the Massachusetts Institute of Technology (MIT) to quantify exposure from supply chain disruptions, taking into account the full end-to-end scope and includes multiple tiers of suppliers into the analysis. Based on estimated “time to recovery,” the method calculates a firm’s financial impact. After a proper risk identification and analysis, companies have to make choices about managing supply chain risks. A common approach towards Supply Chain Risk Management, also called the 4T portfolio approach, is to apply a portfolio of actions to terminate, transfer, treat and tolerate risks (Burtonshaw-Gunn 2008): • • • •

Terminate the risk; avoidance, e.g., by not entering into a contractual relationship. Transfer the risk; move the consequence. Treat the risk; reduce the possibility or mitigate the impact by control measures. Tolerate the risk; accept the possible outcome.

Risks may be mitigated by actions such as e.g. the selection of financially strong suppliers, lean management and six-sigma techniques aimed to reduce or eliminate waste and delays, and the use of visibility tools to track global shipments and take

25 Supply Chain Security

579

Fig. 25.2 The 8-layer supply chain security model (Hintsa 2011)

necessary action if needed, and other approaches to reduce lead times and lead time variation. Other measures include strengthening global logistics competence, predictive modelling, use of air freight, adding inventory, insourcing or near sourcing and (self) insurance. The latter is a way of transferring risk, the cause is not taken away; only the firm’s vulnerability is being managed. Hintsa developed a pragmatic framework to manage supply chain security risks, considering eight different layers (Hintsa 2011) (Fig. 25.2). This model was developed and effectively applied in supply chain security research projects such as CASSANDRA (www.cassandra-project.eu) and CORE (www.coreproject.eu). These European research projects both address the more advanced aspects of supply chain security policies and innovative practices. The aim in both projects is to enhance supply chain security driven by commercial interests, whereas customs and other border agencies can benefit from the advanced control measures implemented in the supply chain and integrate this in their supervision and enforcement strategies. Supply chain visibility is the key enabler for both enhanced supply chain security and commercial propositions, such as supply chain synchronization. Around this idea, CASSANDRA (2010–2014) has developed the concept of data pipelines reusing reliable and high quality source data throughout the supply chain and sharing it with customs. This resulted in promising proof of concepts. In CORE (2014–2018), the overall concept was expanded with the notion of realtime visibility through an event driven architecture and a pull-based mechanism that respects the data governance requirements of data exchange. This also contributes to

580

G. Zomer

the data security requirements necessary to overcome the data sharing reluctance and create trust. Moreover, CORE explores and elaborates the commercial business cases using real-time visibility to enhance supply chain agility and supply chain resilience in ten different demonstrations.

25.2.1 Supply Chain Resilience Christopher and Peck (2004) were the first to define supply chain resilience as ‘the ability of a system to return to its original state or move to a new, more desirable state after being disturbed’. They argue that creating resilient supply chains is needed to effectively manage supply chain risks in the highly complex and interrelated networks of firms contributing to the value chain of products and services. They propose four ways to create supply chain resilience. First, supply chain design objectives should include resilience, next to cost optimization and customer service, for instance by reconsidering the trade-off between efficiency and redundancy in just-in-time and lean management practices. Second, supply chain collaboration and information exchange is essential to reduce uncertainty and enhance visibility of changing risk profiles in upstream and downstream activities. Third, resilience implies agility. Agile capabilities such as supply chain visibility and supply chain velocity (short lead times and synchronized processes) are also resilient capabilities. Fourth, it is important to foster a supply chain risk management culture throughout the organization, similar to the quality culture in Total Quality Management or safety culture within the chemical manufacturing industry. Driven by the global financial crisis in 2008, and the notion of systemic risks, supply chain resilience regained global interest at the World Economic Forum in 2012 (WEF 2012, 2014). It addressed the importance of building a resilience framework to manage systemic supply chain risks, such as oil dependence, information fragmentation, natural disasters, and global economic demand shocks, requiring new mitigation strategies built upon multi-stakeholder collaboration. These mitigation strategies require among others new forms of scenario planning and development of governance models for data sharing and collaboration. These reports also provided the basis for a shift in America’s national supply chain security strategy, reconfirming the importance of building resilience through collaboration; not just within the supply chain, but also between industry and government. The next section elaborates on the challenges that prohibit this from happening.

25.2.2 Conflicting Interests in Control Needs An end-to-end approach in supply chain security is often lacking, it focuses on a collection of individual measures driven by corporate interests and embedded in internal control frameworks (Li 2014). Risk transfer strategies such as insuring financial risks

25 Supply Chain Security

581

and outsource strategies are widely applied. There is not always a sound business case on corporate level to treat or mitigate security risks. Friction costs of organizing and implementing upstream controls and applying gain sharing mechanisms are perceived to be considerable. Consequently, the potential of collaborative chain controls remains underutilized (Zomer et al. 2014). This has also consequences for the alignment of the needs of border agencies and trade and logistics industry. As we see in the next section, supervising authorities deploy joint supply chain security programs that build upon the business efforts to effectively control supply chains against unwanted deliberate acts. But these also have their limitations in the alignment of corporate and societal objectives.

25.3 Control and Governance of Cross-Border Trade (Advanced) Global trade is governed by a variety of rules, regulations and procedures. Many different ministries and bodies are involved in setting those policies and a wide variety of agencies enforcing corresponding controls, such as customs, port health authorities, aviation authorities, coastguard, police, food and product safety authorities, and traffic and transport authorities (Grainger 2008). Grainger described this complex trade governance system from the perspective of UK traders, highlighting the international, EU, US and UK security regimes and linkages between the different regimes and introduced a term for it: the ‘security spaghetti’. The key security programs within these regimes are briefly described in the following sections (Fig. 25.3).

25.3.1 The International Supply Chain Governance Layer Grainger considers several institutional governance layers in supply chain security. The international layer involves institutions such as the International Maritime Organization or the World Customs Organization. They develop programs, policies, standards and recommendations that are transposed into European and national security programs. Some of the most relevant international programs and policies include WCO’s SAFE Framework of Standards, the ISO 28000 standard, the IMO ISPS Code and TAPA’s freight security requirements. The essence of these programs are highlighted below, see also Chap. 7 for an elaborate discussion. (a) World Customs Organization (WCO)—SAFE Framework of Standards The SAFE Framework of Standards is a set of 17 recommendations to customs organizations all over the world to enhance supply chain security and facilitate legitimate trade. It was adopted in 2005 and rests on two pillars; Customs to business partnerships, with the Authorized Economic Operator (AEO) scheme as a successful adopted

582

G. Zomer

Fig. 25.3 Security Spaghetti (Grainger 2007)

worldwide. The AEO scheme is a benefits program for compliant traders offering facilitation in customs controls and simplified procedures. The other pillar—Customs

25 Supply Chain Security

583

to Customs network—includes standards such as harmonization of advance cargo information, introduction of risk management approach and exchange of inspection results of high-risk containers between customs organizations. In Europe, this framework is a guiding instrument in the customs reforms such as the Union Customs Code, the modernization of customs procedures in Europe that entered into force in May 2016. (b) International Standards Organization (ISO)—ISO 28000 ISO 28000 is an international standard for management system for security assurance in the supply chain. It is a voluntary scheme that aims to support businesses to establish, implement, maintain and improve a security management system in conformance with their security policy and accredited by a third party certification body. The certified organizations make a self-declaration of their conformance and must be able to demonstrate this conformance to others. It proposed an integrated risk management approach, embedding security risk management into broader Enterprise Risk Management framework. (c) International Maritime Organization (IMO)—ISPS Code The International Maritime Organization is an agency of the United Nations with responsibility for the safety and security of shipping and the prevention of marine pollution by ships. As a response to the terrorist attacks of 11 September 2001, the IMO responded with the creation of the International Ship and Port Facilities Security Code (ISPS) to protect the maritime sector (including shipping companies, ships and port facilities) from terrorist acts. It requires security threat detection capabilities and preventive control measures need to be taken. Ships must comply to the local security requirements in port areas. Both crew and port staff must be certified being secure. The code was enacted in 2004 in European law by Regulation 725/2004. (d) Transported Asset Protection Association (TAPA)—Freight Security Requirements TAPA is a voluntary association that sets freight security requirements in the global supply chain to combat theft and other losses in international supply chains. TAPA promotes the use of worldwide security standards and best industry practices. With help of high-tech industry experts in security, TAPA has established Security Requirements in Freight, Facilities and in Trucking for manufacturers, suppliers, carriers and logistic service providers. (e) Strategic trade controls and multilateral export control regimes Strategic trade and export controls are elements of compliance with international treaties to prevent proliferation of weapons of mass destruction, limiting access to sensitive technologies that may threaten national security and combat the financing of terrorism. A number of complementary regimes address different security threats, such as chemical and biological weapons (The Australia Group), nuclear weapons (Nuclear

584

G. Zomer

Suppliers Group), delivery systems (Missile Technology Control Regime), and conventional arms and dual-use goods and technologies (The Wassenaar Arrangement). The cost to comply to these regimes is substantial for trade, finance and logistics companies involved in global trade of these goods. It requires a high level of chain control, ensuring that not only a company’s first-tier suppliers comply with the requirements, but also second- and third-tier suppliers. On the demand side, one needs to know the customer (KYC-compliance), but also the ultimate end-user of these goods. In addition, many products, goods and technologies are subject to the dual use regime, it covers about 20% of the total value of traded goods. Moreover, the cost of non-compliance is huge. Companies have bene fined for several hundred millions or even billions of dollars and export stops have even caused bankruptcy of companies.

25.3.2 The European and American Supply Chain Security Governance Layer In addition to the international layer of security programs, Europe has its own supply chain security program. (f) EU Safety and Security measures in the (Union) Customs Code In 2005 and 2006 Europe introduced a number of measures to strengthen supply chain security through adjustments in trade data requirements, better risk-based targeting and faster security checks (EC Regulation 648/2005 and Regulation 1875/2006). From 2009, traders have to submit in an electronic way advanced information on goods entering and leaving the EU, where it is the ultimate responsibility of the carrier, who brings goods into the EU, to lodge this Entry Summary Declaration or ENS. Similar to the AEO concept of the WCO, Europe introduced the Authorized Economic Operator (AEO) concept. This concept involves three types of AEO status: (i) Customs Simplifications; (ii) Security and Safety; and (iii) combined Customs Simplifications/Security and Safety. Benefits include easier admittance to customs simplifications, fewer controls, prior notification and treatment in case of selection for controls. This also has a positive impact on the reliability of the order fulfilment process. Europe is in a transition towards a new framework regulation for customs rules and procedures, called the Union Customs Code (UCC). The UCC will include some adjustments to the pre arrival declaration (called Entry Summary Declaration or ENS) lodging process. It will require some additional data elements and allows for multiple filing of the ENS data elements is allowed by more than one entity. In addition, the UCC introduces an additional requirement for AEO. The AEO needs to demonstrate having practical standards of competence or professional qualifications. Moreover, the use of some simplified or special procedures will only be offered to AEO authorized operators.

25 Supply Chain Security

585

(g) EU Customs Risk Management Action Plan After introduction of the security amendments to the Customs Code, Europe made an evaluation and adopted a policy to strengthen customs risk management with an action plan (COM/2014/527). This plan highlighted actions needed to improve data quality, enhance information sharing amongst customs authorities, avoid duplications of controls and integrate controls better in the supply chains, create a level playing field across the EU in harmonizing the procedures, strengthen interagency cooperation and further develop the partnership with reliable traders. (h) Aviation security Evolving risks and a number of incidents with high visibility have resulted in the need for additional requirements in aviation security. Basic aviation security standards on airport and aircraft security, screening of passengers, cargo, mail and supplies, and staff recruitment and training require additional and or adjusted regulations, such as allowing the introduction of new scanning technologies and additional validation requirements for carriers entering Europe from third countries (ACC3). (i) EU Internal Security Strategy (ISS) In 2010, the internal security strategy for the European Union has been adopted to protect its society against evolving threats, such as organized crime, money laundering, corruption, trafficking and mobile organized crime groups, and cybercrime. Fighting organized crime became a priority because of its disruptive potential for our socio-economic system. Trade in illicit drugs and organized VAT fraud both represent e100 billion per year in Europe. In trade, you see this reflected in stricter anti money laundering regulations and higher fines in case of non-compliance. Another evolving threat is cybercrime. With its potential to disrupt our economic system, it requires extra attention. This was recognized and reflected in the renewed Internal Security Strategy in 2016. Let’s have a look at how new challenges are reflected in policies and procedures in the USA and if there are differences compared to the European approach. (j) The US export control regime and its extraterritorial provisions The United States has its own approach to how to balancing supply chain security and facilitating legitimate trade. Two typical elements include the US export control regime and the multi-layered approach. Export control regimes are set up to control the proliferation of (nuclear) weapons and other sensitive global security items. However, they also offer a tool to protect strategically important technology. Since the USA is the world’s largest producer and exporter of defense items, its policy balances the protection of national security, its commitment to international security and commercial interests. Not only US citizens and US companies are subject to these regulations, but basically anyone dealing with US controlled goods or technology, including non-US based companies with US employees. American jurisdiction is not only being applied to goods and technologies of US origin, but even to products with US technology embedded or product designs

586

G. Zomer

based on US technical data. In a globalized economy, this means that almost all non-US companies involved in the production, trade (financing) or logistics of export controlled goods have to consider the reach of this national policy. This is no sinecure, since the national export control policy in the US is associated with ultra-high costs of non-compliance. For instance, a company re-exporting listed export controlled goods of US origin from Germany would require applying for a license in the USA prior to exporting. And there are three different licensing agencies in three different departments processing over 130,000 licenses and applications in 2009 (Genard 2014). In some cases, exporters were required to apply for multiple licenses from separate departments. It is challenging for companies to comply to such frameworks in an efficient way, it raises complexity and considerable friction costs to global trade of export controlled goods. (k) The multi-layered approach by Customs and Border Protection Customs and Border Protection (CBP) in the United States introduced in 2008 the concept of a layered approach to ensure supply chain integrity. It is built on the following layers to protect against terrorism linked to incoming maritime goods: • Advanced information submission. The 24-hour manifest requires submission of 21 data elements by the carrier 24 hours before loading of the cargo in the port of departure. A supplementary requirement is either Importer Security Filing or 10 + 2 rule, which requires the submission of 10 additional data elements by the shipper, such as the buyer, the seller, the consolidator and the stuffing location. • Targeting and screening of shipment-related information. The Automated Targeting System collects data from different systems including among others import and export declarations, airline reservation data, land border-crossing vehicle data, nonimmigrant entry records, bills, entries, and incident logs. • Partnership programs with the trade industry, such as the Customs Trade Partnership Against Terrorism (C-TPAT). This is a voluntary program offering incentives in inspection processes in return for tight control processes to assure supply chain security. • Partnerships with foreign governments, such as the Container Security Initiative (CSI) and the Secure Freight Initiative (SFI). The CSI program includes prescreening of containers as early in the supply chain as possible, this is being applied in collaboration with oversea customs authorities. The Security Freight Initiative is a scanning project to detect radiation with non-intrusive technology in oversea countries before departure. Many oversea ports participate in the CSI initiative, just a limited number participate in SFI. • Use of Non-Intrusive Inspection (NII) technology for high risk shipments; This mobile gamma-ray imaging technology helps in detecting narcotics, weapons or money hidden between other cargo in containers, trucks or tankers without the need to open them. The Department of Homeland Security presented in 2012 a National Strategy for Global Supply Chain Security. This recognizes the upcoming challenges of dealing with cyber security threats.

25 Supply Chain Security

587

Fig. 25.4 Vision on regulatory supervision of the Customs Administration of the Netherlands, see also Customs NL (2014)

25.3.3 Trusted Trade Lane Supervision (State-of-the-Art) A way to contribute to more effective control and enforcement, whilst reducing the administrative burden for legitimate trade is to apply a system-based approach to regulatory supervision. This considers the entire system of internal controls and releases the focus on individual transactions. It includes a dialogue between companies demonstrating that they are in control and regulatory agencies to assess the effectiveness and adequacy of the controls. As such, customs or other authorities can re-use commercially motivated controls for their own control and supervision purposes. This is called ‘piggy backing’ (Tan et al. 2011). Trusted trader concepts, such as AEO and C-TPAT are based on this principle. However, when the piggy back concept is broadened from internal controls to covering the entire supply chain, we can consider trusted trade lanes. This offers new opportunities for customs and other authorities to strengthen their supervision. The corresponding supervision model applying these principles has been first developed in the CASSANDRA project (Klievink 2014). Dutch customs developed a vision supporting this supervision concept and is the first national customs authority engaged in demonstrating this concept jointly with industry in practice in the CORE project (Fig. 25.4). The trusted trade lane model has four properties: 1. 2. 3. 4.

A supply chain visibility infrastructure, An interface to provide visibility to authorities, Deployment of chain controls to assure trustworthiness, and A chain governance and corresponding business model to incentivize implementation of chain-wide controls.

588

G. Zomer

The chain control measures can be categorized in three clusters: trustworthy partner identification and selection, data quality and data validation controls, and goods and container integrity controls. Below, we present the case description from Royal Flora Holland, which shows how this model has been applied in practice. Case Study: Cut Flowers from Kenya to the Netherlands Consider the transport of cut flowers from Kenya to the Netherlands. Cut flowers are perishable goods and normally transported by plane from Kenya to Europe, but recently sea transport in refrigerated containers is also being explored. Strong variations in temperature have a devastating impact on the product quality, measured in ‘vase life’, the number of days the end consumer can enjoy from a bouquet of roses for instance. So there is a strong commercial incentive not to open the container during travel. Now assume for example that a trustworthy inspection agency, such as the Kenya Plant Protection Agency (KEPHIS) has inspected the cargo before departure. Assume there is a reliable channel to exchange original trade documents about the freight, such as for example the phytosanitary certificate. Also assume, the integrity of the products is safeguarded by a devise that monitors the temperature and is connected to an IT system that generates an alert if strong deviations in temperature occur. Then there is a strong reason for the customs administration to believe that, upon arrival in the Netherlands, the freight is in the same condition as when it was inspected before departure (Zomer et al. 2017). This is exactly what has been demonstrated by Flora Holland, the world’s largest flower auction and cooperation of growers, Dutch Customs, the Dutch Phytosanitary Inspection Authority and the Kenyan Plant Protection Agency. Flora Holland developed and implemented an event-based virtual tracking of shipment journey, using smart devises for end-to-end monitoring of supply chains (Babbler’s). Moreover, they substituted the paper phytosanitary certificate with an e-certificate and made the corresponding data elements available to customs to access together with packing list, pro-forma invoice and entry summary declaration data elements via a pull-driven interface: the ‘customs dashboard interface’. Dutch customs in return can perform in advance risk analysis, while clearance of incoming shipments can be done in parallel with clearance by the phytosanitary authority (sequential processes before). The additional data elements and insight in the chain controls being applied in this trade lane (data quality, partner screening and physical integrity of the goods), result in faster clearance and release.

25 Supply Chain Security

589

25.4 Security in Digitalized Supply Chains and Cyber Resilience (State-of-the-Art) In global trade and logistics, the number of document exchanges related to the international movement of a shipment ranges from 40 to 200. More and more, this document exchange becomes digitalized. Trade facilitation, such as digitalization of Business-to-Government document exchange has a potential to reduce friction costs in global trade with $2.6 trillion. This kind of digitalization is being accelerated by the uptake of single window environments. Also in business-to-business, the transition from paper-based documents towards electronic documents has occurred rapidly, though companies are confronted with significant data interoperability issues and corresponding friction costs. However, despite the merits of digitization in trade and logistics, there is another side of the coin. The fast uptake of digital exchange of status messages and declarations introduces a whole range of new risks from a variety of known and unknown sources and has made the global trade and logistics system vulnerable for cyber threats. Cybersecurity is about protecting IT systems from theft of information, damage to hardware or software or misdirection of its intended services. Cyber events can impact the availability, integrity or confidentiality of IT systems and services. Events can be intentional, such as hacker attacks, but also unintentional, for instance failed software updates. Cyber events can be caused by humans or by nature of a combination of both. Digitized trade and logistics has resulted in an increased reliance on IT systems, further accelerated by the take up of smart devices and associated services (Internet of Things). This has contributed to the growing importance of cyber security and cyber resilience. The cost of cybercrime is considerable, predictions range from $400 billion (CSIS 2014 and Lloyd) to $3 trillion per year (Morgan 2016). And these costs are expected to rise quickly in the coming years. Cyber resilience has a broader scope then cyber security, it not only protects against cyber threats, but anticipates and responds to them in a way to recover quickly and minimize its negative impacts, similar to supply chain resilience. It is based on the same pillars: prepare, protect, detect, respond, and recover; if applied in the context of global supply chains, it is an integral part of supply chain resilience (NIST 2018). But what exactly can companies do to enhance cyber resilience? One may think of firewalls, rigorous passwords, encryption, replacing outdated hardware, and structural testing to improve incident response. In addition, proactive identification of attacks and the creation of a cyber-awareness culture (Chinn 2014) are ways to enhance cyber resilience. Since 2012, supply chain resilience and cyber resilience is one of the key priorities and initiatives (World Economic Forum 2012, 2014, 2017). Many studies highlight the vulnerability of the maritime industry to various cyberrisks and its lack of adequate defenses (CyberKeel 2014). Well-known examples of a cyber-security breach in the global logistics system include the Antwerp terminal hack between 2011 and 2013 and the cyberattack by the Petya virus hitting 17 of the container terminals of APM in June 2017.

590

G. Zomer

Case Study: The Antwerp Terminal Management System Hack A criminal organization used a trade lane from South America to the port of Antwerp to smuggle huge volumes of drugs, hidden as bananas. In order to avoid discovery and inspections, the organization hired professional Belgian hackers to hack the Antwerp terminal systems that manage the transport, storage and release of containers through the port. How did it work? The hackers used a combination of practices: hacking the computer networks of terminals and port companies, distributing malware to enable key-logging, design and configuring of network equipment and placing data tapping equipment after burgling. In doing so, they were able to identify the location where the containers with narcotics were stacked, manipulating the stacking location and even manipulating the release of the containers for hinterland transport and pick up the ‘released’ containers before the real client appeared to collect it. When a container is being released, terminal operators send a pin code to the consignee of the container, after which criminals could obtain this pin code through phishing and hacking. When the security breach was exposed, the port installed a firewall. But this did not stop the criminals from operating this smuggle trade lane; they penetrated physically into the port and installed wireless bridges on the operating computers, opening a direct access to the operating system. Highly professional IT hackers tapped the complete data traffic of companies like DP World, PSA, MSC, Katoennatie and CSAV. And IMSI catchers—to be used by police and national security officers only—were in possession of this gang and were used to interrupt mobile GSM communication, even from police and customs officers. This practice continued for over two years, before the port discovered the cybercrime (Pol 2015). Cyber security and cyber resilience gained a lot of attention and now have high priority within ports. The Port of Rotterdam appointed in June 2016 a Port Cyber Resilience Officer. The Port CRO is responsible for realizing the Port CRO pro-gram. The purpose of this program is to collectively increase the cyber resilience in the port, raise cyber security awareness, intensify the preparedness of the different organizations in the port environment (including inspection and control authorities), and strengthen risk management in the particular domain of cyber threats.

The case example and more recent Petya virus attack of APM terminals highlight that cyber resilience has the character of a rat race. Attackers every day try to find weaknesses and ways to exploit these weaknesses. Therefore, it will remain one of the key priorities in supply chain risk management the coming years.

25 Supply Chain Security

591

25.5 Further Reading A nice introduction to the topic of supply chain resilience is provided by Christopher (2004) who proposes several ways to build resilience capabilities in supply chains. Hintsa (2011) presents a comprehensive compendium of relevant supply chain risk analysis theory, while Hintsa and Uronen (2012) extensively document key supply chain security policies, regulations, standards and guidebooks. Primary research carried out in over 200 organizations on ‘game changing’ actions for business and governments is reviewed in Kaplan et al. (2015). In Tan et al. (2011) a collection of papers is presented on how IT innovation can be applied to enhance supply chain security and facilitate trade. The case on the Antwerp Terminal System Management hack has been described by Van de Pol (2015).

References Burtonshaw-Gunn SA (2008) The essential management toolbox: tools, models and notes for managers and consultants. Wiley & Sons. ISBN: 978-0-470-51837-3 Center for Strategic and International Studies (CSIS) (2014) Net losses: estimating the global cost of cybercrime, June 2014 Chinn D, Kaplan J, Weinberg A (2014) Risk and responsibility in a hyperconnected world: implications for enterprises, January 2014 Christopher M, Peck H (2004) Building the resilient supply chain. Int J Logist Manag 15(2):1–14 Cranfield University (2002) Supply chain vulnerability. Executive summary, Cranfield School of Management. http://www.som.cranfield.ac.uk/som/dinamic-content/research/lscm/download s/Vulnerability_report.pdf Customs NL (2014) Pushing boundaries: the customs administration of the Netherlands’ point on the horizon for the enforcement on continuously increasing flows of goods. Customs Administration of the Netherlands, The Hague Cyberkeel (2014) Maritime cyber risks. https://maritimecyprus.files.wordpress.com/2015/06/mari time-cyber-risks.pdf Dittmann JP (2014) Managing risk in the global supply chain. In: Game changing trends in the supply chain. Supply Chain Institute, University of Tennessee Grainger A (2007) Supply chain security: adding to a complex operational and institutional environment. World Cust J 1(2) Grainger A (2008) Trade facilitation and import-export procedures in the EU: striking the right balance for international trade. Briefing Paper requested by the European Parliament’s Committee on International Trade, March, Brussels Genard Q (2014) European responses to extraterritorial claims by the United States: lessons from trade control cases. In: Non-Proliferation Papers, No 36 Gupta S (2015) Study on the concept of supply chain security. Glob J Res Anal 4(7). ISSN No 2277-8160 Hintsa J (2011) Post-2001 supply chain security—impacts on the private sector. University of Lausanne: Faculty of Business and Economics, Doctoral thesis, Lausanne Hintsa J, Uronen K (eds) (2012) CASSANDRA Compendium, accessible as Deliverable Cassandra D1.1. www.cassandra-project.eu/downloads Kaplan JM, Bailey T, O’Halloran D, Marcus A, Rezek C (2015) Beyond cybersecurity: protecting your digital business. Wiley Publishing Co

592

G. Zomer

Li TS (2014) Establishing an integrated framework for security capability development in a supply chain. Int J Logist Res Appl 17(4) Malenstein J, Schewe W, Zomer G, Klievink B, Nijdam M, Visscher W (2014) CASSANDRA final protocol, accessible as Deliverable Cassandra D6.3. www.cassandra-project.eu/downloads Morgan S (2016) Hackerpocalypse: a cybercrime revelation NIST (2018) Framework for improving critical infrastructure cybersecurity, Version 1.1, April 2018. https://www.nist.gov/sites/default/files/documents/cyberframework/cybersecurity-f ramework-021214.pdf Tan YH, Bjørn-Andersen N, Klein S, Rukanova B (eds) (2011) Accelerating global supply chains with IT-innovation. Springer, Berlin Thomas AR (2010) Supply chain security—international practices, and innovations in moving goods safely and efficiently Talbot J, Jakeman M (2009) Security risk management body of knowledge. Wiley Van de Pol W (2015) Gehackte haven, Cokesmokkel 2.0. In: Crimesite.nl. http://www.crimesite.n l/gehackte-haven-cokesmokkel-2-0-1/ Voster W, De Bruijn J (eds) (2016) Cyber security supply chain risk analysis. Research report of Royal Dutch Shell, Nederlandse Gasunie, Nuon, TenneT, Alliander, with support of the Dutch National Cyber Security Centre. https://www.cybersecurityraad.nl/binaries/Brochure%20Cybe r%20Security_ENG_WEB_tcm56-79500.PDF World Economic Forum (2012) Risk and responsibility in a hyperconnected world—pathways to global cyber resilience World Economic Forum (2014) Risk and responsibility in a hyperconnected world—insight report World Economic Forum (2017) Advancing cyber resilience principles and tools for boards Zomer G, De Putter J, Van Oosterhout M, Van Dalen J, Mueller R, Barz A (2014) Integrated evaluation report of CASSANDRA—common assessment and analysis of risk in global supply chains, accessible as Cassandra Deliverable D5.2. www.cassandra-project.eu/downloads Zomer G et al (2017) CORE deliverable D2.4 supply chain security controls 2 Zsidisin GA (2005) Approaches for managing price volatility. In: 5th international research seminar on risk and the supply chain. Centre for Logistics and Supply Chain Management, School of Management, Cranfield University

Chapter 26

Trends in E-commerce, Logistics and Supply Chain Management Gregor Sandhaus

Abstract Over the last decade e-commerce has grown to a multi-billion business. “Online-shopping” and “Electronic payment” are typical expressions that are associated with e-commerce but they only describe the non-material communication between buyer and seller. The manufacturing and transportation process from raw materials in production facilities to a finished product at the customer location cannot be accomplished via the internet. It requires the solid handling of goods. E-commerce technologies can support the administration that goes alongside with this process but they do not replace the required physical treatment. This chapter takes a look the interdependencies of logistics, supply chain management and e-commerce. The first section explains how e-commerce, supply chain management and logistics are linked whereas the second section provides an overview of advanced technologies that enable the transportation of goods along the supply chain. A selection of state-of-the-art business and technology trends is presented in the third section.

26.1 Basic Elements of E-commerce Systems, Supply Chain Management and Logistics Fulfilment “Since it began in 1995, global e-commerce has grown from a standing start to a e1.2 trillion retail, travel, and media business and a e12.4 trillion business-to-business juggernaut, bringing about enormous business change in firms, markets and customer behavior.” With this sentence, Laudon and Traver (2015) underlined the importance of e-commerce in 2015. Electronic commerce refers to a wide range of online business activities for products and services. It includes business transactions in which the parties interact electronically rather than by physical exchanges or direct physical contact. E-commerce is usually associated with buying and selling via the Internet, or conducting any G. Sandhaus (B) FOM University of Applied Sciences, Mettmann, Germany e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_26

593

594

G. Sandhaus

transaction involving the transfer of ownership or rights to use goods or services through a computer-mediated network (Andam 2016). Logistics is a crucial backbone of e-commerce transactions. Without a superb logistics, the e-commerce business would never have experienced the exponential growth as we have witnessed today (Chap. 2, Sect. 2.3.3). Here, logistics indeed includes not only all physical transfers but also the electronic ordering, billing and tracking/tracing systems that are in place nowadays (Cerasis 2015). Logistic service providers are aiming at sending high numbers of items (e.g. packed on one pallet) within one delivery. Because of economies-of-scale arguments, the transport costs can become relatively small. However, today’s online shopping tears this pallet down into many shipping units. Each unit is then individually sent to the customer. Therefore, in the B2C-sector (business-to-consumer) the increasing number of parcels delivered by courier, express and parcel (CEP) service providers leads to an increase of transport costs in relation to the order volume (Arnold et al. 2008). While B2C e-commerce gets the most media attention, B2B (business-tobusiness) e-commerce revenue is globally twice as high as that of B2C. The entry of e-commerce giants such as Alibaba and Amazon into B2B has accelerated the trend of B2B websites becoming more like their B2C counterparts. Online B2B sellers recognize that the customer experience in a B2B environment is just as important as the customer experience of B2C (Kaplan 2016). Demand-oriented manufacturing and individualized products and services cause smaller but more frequent transport orders. Therefore, also in the B2B-sector transport costs increase in relation to the order volume. While e-commerce has many advantages in terms of overcoming geographical limitations, gaining new customers, creating markets for niche products, remaining open all the time etc., logistics is one of the areas where complexity increases as one moves from physical retail to e-commerce. In a retail store, one can simply hand over the merchandise to the customer. In e-commerce issues of warehousing, inventory, packing, shipping, and tracking, still need to be taken care of (Khurana 2016). Therefore, all logistic related elements alongside the supply chain remain relevant for e-commerce. Today’s supply chain management systems support logistic processes like inbound logistics, warehousing and production, outbound logistics and return logistics as shown in Fig. 26.1. The Supply Chain Operations Reference Model (SCOR) in Fig. 26.1 displays a supply chain as a string of processes for procuring raw materials, transforming these materials into intermediate and finished products, and distributing the finished products to customers (see also Chap. 3). It links suppliers, manufacturing plants, distribution centers, retail outlets, and customers. Materials, information, and payments flow through the supply chain in both directions (Laudon and Laudon 2014). Supply chain systems enable companies to model their existing supply chain, generate demand forecasts for products, and develop optimal sourcing and manufacturing plans. This helps companies to

26 Trends in E-commerce, Logistics and Supply Chain Management

595

Fig. 26.1 Logistic fulfillment in SCOR (cf. Chap. 3)

• determine the optimum production output (cf. Chaps. 12 and 19), • to establish appropriate inventory levels (cf. Chap. 20) for raw materials, intermediate products, and finished goods (warehousing), and • to plan the transportation of incoming goods and material (inbound logistics, cf. Chap. 11) and the product delivery (outbound logistics, cf. Chap. 14). E-commerce helps companies to enter international markets, to outsource manufacturing operations, and to obtain supplies from other countries. Supply chains have become international with additional complexities and challenges. Global supply chains typically span greater geographical distances and time zones and have participants from different countries. Supply chain management, for example, may need to reflect foreign government regulations and cultural differences (Laudon and Laudon 2014). Considering that most companies have different suppliers and distribution partners, today’s supply chains may better be seen as a network of organizations and business processes as shown in Fig. 26.2. These networks of manufacturers, logistics suppliers, outsourced manufacturers, retailers, and distributors communicate in many directions simultaneously. Participants exchange information about what to produce, store, and move. They match supply to demand, reduce inventory levels, improve delivery service, speed up product time to market, and use assets more effectively. In some industries, supply chain costs add up to 75% of the total operating budget and therefore efficient supply chain management can have a major impact on companies’ profit (Laudon and Laudon 2014). To highlight the many aspects of e-commerce, the following case study of Dell Computer Corporation has been presented by Turban and King (2012a, b).

596

G. Sandhaus

Fig. 26.2 Supply chain network with multidirectional communication

26.1.1 Case Study: Dell The Problem Founded in 1985 by Michael Dell, Dell Computer Corporation was the first company to offer personal computers (PCs) via mail order. Dell designed its own PC system and allowed customers to configure their own customized systems using the build-to-order concept. This concept was, and is still, Dell’s cornerstone business model. By 1993, Dell had become one of the top five computer makers worldwide, threatening Compaq, which started a price war. At that time, Dell was taking orders by fax and letter post and losing money. Losses reached over $100 million by 1994. The Solution Direct Marketing Online: The Internet in the early 1990s provided Dell with an opportunity to expand rapidly. Dell implemented aggressive online order intake and opened subsidiaries in Europe and Asia. Dell also started to offer additional products on its website. This enabled Dell to battle Compaq, and in the year 2000, Dell became number one in worldwide PC shipments. Direct online marketing is Dell’s major electronic commerce activity. Dell sells to the following groups:

26 Trends in E-commerce, Logistics and Supply Chain Management

• • • •

597

Individuals for their homes and home offices Small businesses (up to 200 employees) Medium and large businesses (over 200 employees) Government, education, and health care organizations.

Sales to the first group are classified as B2C. Sales to the other three groups are classified as B2B. In addition, Dell sells refurbished Dell computers and other products at electronic market places. In 2006, Dell opened physical stores, mainly in reaction to customer demands. Business-to-Business: Most of Dell’s sales are to businesses. Whereas B2C sales are facilitated by standard shopping aids (e.g. catalogs, shopping carts, credit card payments) B2B customers obtain additional help from Dell. Dell provides each of its nearly 100,000 business customers with Premier Dell service. This allows customers to browse, buy, and track orders on a Dell website customized for the user’s requirements and enables authorized users to select preconfigured PCs for their business unit or department. A more advanced version, Premier B2B, provides automatic requisition and order fulfillment once an authorized user has chosen to buy a PC from Dell. This allows authorized staff in their intranet to purchase PCs through a portal that connects directly to Dell’s systems. In addition to supporting its business customers with eprocurement tools, Dell also uses e-commerce in its own procurement through electronic tendering when it buys the components for its products. E-collaboration: Dell has many business partners with whom it needs to communicate and collaborate. For example, Dell uses shippers, such as UPS and FedEx, to deliver its computers to individuals. It also uses third-party logistics companies to collect, maintain, and deliver components from its suppliers, and it has many other partners. Dell is using e-commerce technologies to facilitate communication and reduce inventories. This technology links customers’ existing enterprise resource planning (ERP) or procurement systems directly with Dell and other trading partners. Dell also educates customers in its technologies and offers suggestions on how to use them. Finally, Dell has a communication system with its over 15,000 service providers around the globe. E-customer Service: Dell uses a number of different tools to provide customer service. To leverage customer relationship management (CRM) Dell provides a virtual help desk for self-diagnosis and service as well as direct access to technical support data. In addition, a phone-based help desk is open 24/7. Customers can also arrange for a live chat with a customer care agent. Product support includes troubleshooting, user guides, upgrades, downloads, news and press releases, FAQs, order status information, a “my account” page, a community forum (to exchange ideas, information, and experiences), bulletin boards and other customer-to-customer interaction features, training books, and much more. Dell also offers educational programs at learndell.com. Dell

598

G. Sandhaus

keeps a large customer database. Using data mining tools, it learns a great deal about its customers and attempts to make them happy. The database is used to improve marketing as well. Intrabusiness E-commerce: To support its build-to-order capabilities, to improve its demand-planning and factory execution accuracy, to reduce orderto-delivery time, and to enhance customer service, Dell partnered with Accenture to create a new, high-performance supply chain planning solution. This solution paid for itself five times over during the first 12 months of operation, enables Dell to adapt more quickly to changing technologies and the business environment, maintaining its position as a high-performance business. Dell also has automated its factory scheduling, demand-planning capabilities, and inventory management using information technology and e-supply chain models. The Results Dell has been one of Fortune’s top five “Most Admired” companies since 1999, and it continuously advances in the rankings of the Fortune 500 and the Fortune Global 500. Dell has over 100 country-oriented websites, and profits are nearing $3 billion a year. Dell also is expanding its business not only in the computer industry but also in consumer electronics. It is clearly an example of e-commerce success.

26.2 Advanced Technologies for Online and Stationary Commerce Including Back-End Logistics Processes No e-commerce without logistics! Whenever a consumer orders a product online, that product has to be shipped from one place (a warehouse or fulfilment center) to another (the consumer’s house, a pick-up point, et cetera). Hence, logistics plays a very important role in the entire e-commerce process. In order to achieve success in the online retail industry, one has to have a very solid logistics plan in place (Chap. 2, Sect. 2.3.3). Traditional distribution to retail stores and distribution centers via pallet loads might not be appropriate for e-commerce. More variable but smaller lot sizes turn logistics from a bulk into a bit by bit business alongside the supply chain. The increasing number of logistic processes requires efficient, consistent and cost-effective handling for each logistic step. In the following paragraphs, some of today’s technical concepts of logistic processes are briefly explained. Warehouse Management System Integration of Third-Party Logistic Providers with ERP As an alternative to in-house fulfillment, a large number of retailers have outsourced their e-commerce order processing to third-party logistics (3PL) providers that are

26 Trends in E-commerce, Logistics and Supply Chain Management

599

Fig. 26.3 WMS/ERP data exchange (Realisable 2013)

equipped and experienced to handle typical online needs. Retailers whose online orders are growing at an accelerated rate often prefer the flexibility of working with a 3PL provider that can help them continue to expand. In addition, 3PL providers are also gaining acceptance because of the sophisticated software engines that link their warehouse management systems (WMS) with the ERP software of the retailer (Grafes 2013). As shown in Fig. 26.3, this link allows 3PL providers to initiate pick orders almost immediately after receiving the corresponding sales order from the ERP system. Similarly, goods received are synchronized with the underlying purchase. Fast Delivery Particularly in areas of high population density e-commerce retailers offer fast delivery services. India’s largest online marketplace Snapdeal.com, for example, has launched a new service that allows the delivery of packages to their customers within one hour after placing the order (Economic Times 2015). Technical concepts for fast delivery are 1. Traffic jam forecasting: Based upon historic data navigation systems can estimate the average traffic congestion and therefore suggest alternative routes rather than guiding the driver into an upcoming traffic jam. 2. Anticipatory shipping: To determine which products are in demand in which area Amazon invented an algorithm that examines user behavior (e.g. time on site, shopping cart activity, wish list, duration of views) on an e-commerce platform. This algorithm is used to predict the purchase of a product so that the fulfillment of the order can be executed in shorter time (Ulanoff 2014). 3. Picking optimization: Real-time forwarding of purchase orders from the e-commerce system to the WMS and efficient picking and packing are also crucial for fast delivery. To accelerate order picking in the warehouse intelligent software systems have been deployed either to route storekeepers efficiently through the shelfs (pick by light, pick by voice) or to automate the picking process with Auto-

600

G. Sandhaus

Fig. 26.4 Worldwide annual supply of industrial robots

matic Storage and Retrieval Systems. Depending on measure, weight, sensitivity and variance automated packing systems can be deployed to increase processing speed in the warehouse. Data Analytics within Manufacturing One aspect of the “Big Data” trend is to deploy data in the manufacturing process. Today manufacturers improve their inventory based upon the information presented by both supply chain and operations data. Logistics and Transportation Managers have reports available to reduce transportation costs—a heavy portion of the expense budget as described earlier in this chapter, see also Chap. 21. Based upon historic data and modern, sensor-based condition monitoring techniques manufacturers can predict when a machine on the production floor is going to fail so they can perform preventative maintenance before a failure causes expensive unscheduled downtime (Gillman 2014, see also Chap. 22 of this volume). Predictive analytics helps manufacturers to understand the changes in demand for their products so that they can determine how to allocate their resources. “Good predictive analytics modelers find additional factors that influenced sales in the past and apply those factors into forecasted sales models” (Gillman 2014). Robotics In 2015, robot sales increased by 15% to 253,748 units, again by far the highest level ever recorded in one year. The main driver of the growth was the electronics industry, metal industry and the chemical, plastics and rubber industry. As shown in Fig. 26.4, the demand for industrial robots since 2010 has increased considerably due to the ongoing trend towards automation and the continued innovative technical improvements in industrial robots (International Federation of Robotics 2016) The International Federation of Robotics (IFR) names the following reasons why there is such an adoption of robotics:

26 Trends in E-commerce, Logistics and Supply Chain Management

601

• Industry 4.0, linking the real-life factory with virtual reality, will play an increasingly important role in global manufacturing. • Human-robot collaboration will have a breakthrough in this period. • Compact and easy-to-use collaborative robots will drive the market in the coming years. • Global competition requires continued modernization of production facilities. • Energy-efficiency and the use of new materials, e.g. carbon-composites, require continued retooling of production. • Growing consumer markets require expansion of production capacities. • Decline in products’ life cycles and an increase in the variety of products require flexible automation. • Continuous quality improvement requires sophisticated high tech robot systems. • Robots improve the quality of work by taking over dangerous, tedious and dirty jobs that are not possible or safe for humans to perform. • Small and medium sized companies will increasingly use industrial robots.

26.3 State of the Art in E-commerce, Logistics and Supply Chain Management Historically the focus in retail was to scale up efficiently, with a focus on cases, pallets, truckloads, container ships and cargo. However, with the emergence of “buy online” and “ship from store” retailers are paying more attention to building economies of scale with small transactions. To master this, two major developments in retail were identified in the previous two chapters: 1. The network of SCM is enhancing in terms of geographical expansion and accurate control of business processes. 2. A detailed insight into available data enables the optimization and automation of manufacturing and transportation. When extrapolating these developments into the next 5 years the following trends for retail and logistics will almost surely be visible. Warehouse Automation and Robots While the IFR names several reasons for robotics in manufacturing, also WMSs have been experiencing robust growth regarding warehouse automation. Companies have invested in goods-to-person automation to effectively meet the requirement of online-order fulfillment like increased agility and responsiveness. Shuttle systems are continually maturing, become more agile and are increasingly being adopted. Mobile autonomous robotics (e.g. Kiva Systems) have been used to support these processes. Since then, a number of start-ups and large companies like Amazon and Google have been investing in mobile autonomous systems to meet these fulfillment process requirements (Banker et al. 2016).

602

G. Sandhaus

Distribution Centers Even though the above-mentioned warehouse automation can reduce manual labor, more employees are required to handle e-commerce related tasks in distribution centers. Particularly the return of goods, same day delivery and small orders are labor intensive because they need to be manually sorted and processed. Therefore, operators of distribution centers occupy up to twelve times more employees in e-commerce distribution centers compared to traditional hubs (Buckley 2016). In addition, the distribution center requirements depend on the product portfolio of an e-commerce operator and can strongly vary. Seasonal fluctuations, type of goods (e.g. groceries), proximity to customers, available employees lead to more complex business models of distribution centers. On the other hand, the expanding online business can also lead to a higher demand of distribution centers (Buckley 2016). Information Sharing To reach as many customers as possible retailers often choose an omnichannel sales strategy. With this strategy, manufacturers and retailers do not own the complete supply channel. Information sharing becomes crucial to understand customer needs in business areas of accelerated product launches, customized product lines and more frequent shipments (e.g. in the fashion industry). Data sharing on trends such as “what’s hot” and “what’s not” enable both parties to work together more efficiently (Doyle 2016). E-commerce sites, for example, require product information in a different data format than provided by the manufacturer. To bridge this technical gap, data transformation software exists. This information also needs to be shared with CRM, order management, ERP, WMS and product information management systems. According to Gartner, e-commerce sites have to integrate technology from 15 or more vendors and therefore efficient information sharing becomes a competitive advantage (Lamont 2016). Return Logistics The buy anywhere, deliver anywhere philosophy of omnichannel retailing leads to a customer expectation to also return anywhere—in person at the store or via package delivery to a distribution center. This requires a level of return process sophistication in terms of flexibility and technology that is not fully developed, yet (Gibson et al. 2016). This lack of sophistication is currently not crucial for e-commerce. According to an online survey by the E-Commerce Center in Cologne, Germany, “fashion and accessories” have the highest return rate. However, even in this segment 46% of the respondents of the survey announced a return rate of less than 20%. On the other hand, the authors of the survey concluded that frequent e-commerce shoppers have higher expectations regarding the return processes. Assuming that users shop more and more online return logistics can become a competitive advantage in the e-commerce business (Brusch and Eva 2013). There is also a relationship between the method of payment and the percentage of articles that are returned. Customers that receive their orders before the actual payment (for instance, when the bill is enclosed in the shipment) are more likely

26 Trends in E-commerce, Logistics and Supply Chain Management

603

to return the shipment than customers that have prepaid the order (Morganti et al. 2014). How companies handle their returns is a reflection of their business that customers will remember. It’s a mark of customer service. Improvements to returns handling technology and Return Merchandise Authorization will make it easier to track a return, with more data available and visible to the customer (Longman 2016). Shortage of Employees In the next 10 years, the US Census Bureau estimates that the population of people of 65 and older will increase by 37.8%. The population of those between 18 and 64 will rise by only 3.2% while those between 18 and 24 will actually decrease in number (USA Today, May 7, 2015). Similar numbers are reported for Europe (see e.g. Federal Ministry of the Interior of Germany 2011). Without immigration the expected shortage of employees will lead to higher wages and therefore to higher prices. The growing scarcity of hourly workers affects retailers across the entire supply chain, but the areas of distribution center labor, truck drivers, and store associates are particularly affected. While the baby boomer generation is in the process of retiring from the workforce, no follow-up generation is in place. With the Millennial generation attending college and earning university degrees in greater proportion than any earlier generation, a dearth of younger talent willing to work in hourly labor positions can be expected (Council of Economic Advisors 2014). Another casualty of the Baby Boomer retirement trend is the possible loss of senior supply chain leaders. With smaller subsequent generations, the population of promotion-ready SCM talent is thin. Part of this is due to the flattening of managerial levels with fewer opportunities to develop cross-functional skills. In addition, the talent being developed by SCM programs in universities will not be ready for top leadership positions for years to come. Retailers are being forced to fill their SCM leadership talent gap in novel ways. For example, they might incentivize existing leaders to remain in their roles longer, re-design roles in order to slow down the pace of leaders’ departure, e.g. by defining part-time roles, or encourage them to remain attached to the organization as consultants or mentors for a period of time. This could mitigate rapid brain drain and support new SCM leaders to expand their roles (Gibson et al. 2016). Cloud Applications for Logistics Similar to other industries, cloud applications are also transforming supply chain logistics. Currently, about one quarter of all enterprise applications are operating within the cloud. According to Oracle CEO Mark Hurd, by 2025 80% of all production applications will be in the cloud (Harper 2015). Whether you believe this prediction or not, an application change within supply chain logistics can be observed. Because cloud applications can offer sophisticated capabilities to small and medium businesses, transportation management systems and order management systems are gaining more and more market share as cloud applications. Why is supply chain logistics moving to the cloud? There are mainly two reasons (Carr 2016): 1. Easy setup makes it simple to equip the mobile workforce of supply chain logistics with mobile applications.

604

G. Sandhaus

2. Low startup costs reduce risk and allow logistics departments and companies to take advantage of new applications without the cost and IT footprint associated with on premise solutions. As mentioned in the subsection “information sharing”, there is a demand for exchanging data between manufacturers and retailers. Cloud applications ease this data exchange and therefore support the collaboration. Collaboration Information sharing due to an omnichannel sales strategy is only one reason for companies to innovate their supply chains. In addition, partnerships and collaborative relationships are often seen as major success factors for a sustainable and efficient supply chain network (Wessel and Greenway 2014). Collaboration is here defined as “the management of upstream and downstream relationships with suppliers, distributors and customers to achieve greater customer value-added at less total cost.” Both a major challenge and an opportunity is the collaboration between competitors. Examples of this “co-opetition” have been successful in many industries. For example, carmakers Ford and Volkswagen Group co-developed and manufactured the Ford Galaxy, Seat Alhambra and Volkswagen Sharan. The advantage of this co-opetition is that a category or market can be developed at lower risk to each organization. In logistics, co-opetition could be utilized to lower CO2 and reduce costs by joint transports. For a successful collaboration, some prerequisites have to be in place (Wessel and Greenway 2014): 1. Both organizations need a common focus/commitment to a basic strategy. For example, in the retail supply chain this could be a focus on the customer. 2. Because information sharing (see above) becomes critical, IT systems need to be in place to enable this. 3. An agreed joint process is required. In practice, this often presents a problem, as it is not uncommon for organizations to have little understanding of their own internal processes. So agreeing on a joint one can be difficult. 4. Both parties should be measuring the success of the relationship in a common way using the same measures. Only if there are advantages for each party there is also a motivation for collaboration. Internet of Things To provide highly integrated “Transportation and Warehouse Management Solutions” connecting in-vehicle sensors and other integrated devices over the network the term “Internet of Things” (IoT) has developed to one of the most commonly used terms in logistics. Radio Frequency Identification (RFID) transponders with embedded sensors in transport vehicles or containers will continuously capture data and share it with logistic applications (see also Chap. 30). The IoT will find many implementations in logistics, from a retailer who needs to monitor the quality of food being shipped, to a clinic that needs to control temperature of the box transporting a specimen (Bhaskar 2016). Future large-scale IoT deployments might be (Heutger and Kückelhaus 2016):

26 Trends in E-commerce, Logistics and Supply Chain Management

605

Fig. 26.5 Business (orders or their gross merchandise volume) from mobile (Meeker 2015)

1. Connected warehouses that increase the transparency and localization of all assets through the tagging of individual items, pallets, and operational hardware. These smart objects can transmit information about their current order, content, and location. This enables automated inventory management with real-time visibility on inventory levels and item conditions. 2. Intelligent transportation solutions that increase transparency and integrity in the supply chain through innovative smart truck concepts. For example, in-vehicle telematics can collect data on movements and idle time to maximize fleet and asset utilization. 3. The connected consumer and the proliferation of smart products and home appliances (e.g., smart locks) that could enable new delivery concepts, such as an automatic replenishment service (e.g., a grocery order can be triggered by a smart fridge) or secured in-home delivery services. This can offer more visibility to the consumer and helps to avoid unnecessary collections. Mobile Shopping According to Mary Meeker’s 2015 report on internet trends, mobile e-commerce plays a significant role. As shown in Fig. 26.5 the share of transactions performed through mobile devices has roughly quadruplicated between 2013 and 2015. Alibaba, eBay, Snapdeal and other online shopping websites report that up to 70% of their online orders or their gross merchandise volume share are initiated through mobile devices (Constine and Escher 2015). Online Grocery Singh and Aditi (2016) identify online grocery as the strongest e-commerce industry in India and name Bigbasket and Localbanya as two of the big brands in this category. This vertical however is currently hardly profitable. Sinha and Paul (2015) identified

606

G. Sandhaus

delivery expenses, warehouse picking labor and occupancy as the three major cost drivers that eat up the relatively small margins of the food industry. Among seven e-commerce models they identified “in-vehicle pickup” (customer stays in car while an employee associate takes order to the car) and “van delivery from distribution centers” as the most cost-efficient delivery systems at present. While shoppers are willing to pay for the convenience of these services, fast delivery (less than one day) seems to be the main obstacle for ordering grocery online. Due to economy of scale and supply chain innovations, Sinha and Paul (2015) predict a 1% annual cost reduction rate. Assuming a 15% annual growth rate in volume e-commerce for groceries can reach profit margins of 10–12%. Big Data According to McAfee et al. (2012) companies that characterized themselves as datadriven are on average 5% more productive and 6% more profitable than their competitors. Big data analytics enables retailers to use existing data more efficiently, drive a higher conversion, improve decision-making and innovate new business models in products and services (Miller 2012). Akter and Wamba (2016) state that big data analytics could allow e-commerce companies in particular to achieve a range of benefits such as • enhanced pricing strategies for products and services, • targeted advertising, • better communication between research and development and product development, • improved customer service, • improved multi-channel integration and coordination, • enhanced global sourcing from multiple business units and locations and • suggesting models and ways to capture greater business value. In the same study, they perform a thorough literature review to explore big data in e-commerce environments. The key findings are summarized below. 1. E-commerce decision makers need to address various business challenges. Some of them are: a. How do processes of big data analytics influence each other and what are their joint effects on a company’s performance? b. What training is required so that employees at all levels can deal with big data? c. What legal prerequisites need to be met to allow the use of big data? d. How can big data improve operations, quality management, and supply chain management? 2. Big data has to be understandable and trustworthy to employees. To gain acceptance, managers should present big data in an understandable format, e.g. through a dashboard, reports or a visualization system.

26 Trends in E-commerce, Logistics and Supply Chain Management

607

3. Particularly e-commerce companies are faced with huge amounts of data originating from different channels. Finding the right information about each customer from this large amount of data is very challenging. At the same time, this challenge is also a huge opportunity to create personalized offers, set dynamic prices, and use the right channels to provide consumer value. 4. Big data bears the risk of redundant, inaccurate, and duplicate data, which might undermine decision-making processes. Highly sophisticated analytics is meaningless if inappropriate data are in place or poor quality data are used. 5. Data security and authorized usage of big data are self-evident but expensive prerequisites. Even though consumers increasingly share personal information in e-commerce sites or in social networks, it is expected for companies to protect consumers’ privacy. 6. Technical, analytical, and governance skills are required to run a big data environment. Data scientists should have communication skills and understand the business of the e-commerce company. 7. The business contribution of big data has to be clear and measurable. Environmental Implications of E-commerce Due to increased environmental concerns in recent years, sustainability, protection of the environment, carbon footprint and green logistics are key words that have been attracting more attention from researchers and practitioners, see also Chaps. 9 and 29, or Mollenkopf et al. (2010). Focusing on e-commerce, the • • • •

usage of information technology, redesign or use of additional packaging, physical distribution of items, logistics activities such as transportation and warehousing

represent key components of the environmental sustainability of the entire supply (McKinnon et al. 2010). Mangiaracina et al. (2015) examined 56 papers that were published in 38 different scientific journals regarding their research methodology and content. Papers, that were more focused on how e-commerce logistics can contribute to sustainability, identified mainly transportation planning and management but also warehousing, packaging, and distribution network design as their interest of research. Within transportation activities subtopics as inefficient deliveries, usage of low-carbon-emission vehicles, overnight deliveries, failed deliveries, and returns have been identified. Papers, that were more focused on the environmental effects of e-commerce where further classified by the applied indicators (i.e. energy use, gas emissions, waste generated, and traffic mileage) in which more than 90% of the examined documents were assigned to the categories “gas emissions” and “energy use”. Mangiaracina et al. (2015) suggest four areas for future research: 1. Vertical focus on clothing and electronics: The majority of the examined publications refer to studies in the book and grocery industry. Although clothing and electronics had the highest growth in sales over the last five years these verticals have hardly been considered in scientific research.

608

G. Sandhaus

2. Omnichannel strategies: Environmental implications and impacts of multichannel shopping experiences have not yet been investigated in depth. 3. Measuring the environmental impact of e-commerce alongside the supply chain: The majority of papers compares e-commerce with traditional retailers so far. 4. More focus on transportation planning and distribution network design, as these areas are believed to have the highest environmental impact.

26.4 Further Reading In this chapter, only some state-of-the-art trends in e-commerce and logistics have been briefly explained. There are more business and technology trends, which affect the supply chain, and are worth mentioning, e.g. sharing economy, green logistics, 3D printing, augmented reality, etc. These trends are discussed in several chapters of this volume. Some of today’s SCM applications have many of the above-mentioned technologies already deployed—at least to some rudimentary level. For example, Cloud Logistics (www.gocloudlogistics.com) offers solutions for SCM and TMS as a cloud service and has integrated interfaces to logistic service providers. Also the business challenges—e.g. information sharing and return logistics—have already been addressed. So, if someone came up with the question “Are there any really NEW trends?” the answer would be a clear “No”. Operations, logistics and supply chain management are continuously evolving and the trends mentioned above had—and still have—a major impact on this evolution. They all pursue one goal: To satisfy customer needs.

References Akter S, Wamba SF (2016) Big data analytics in e-commerce: a systematic review and agenda for future research. Electron Markets 2:173. https://doi.org/10.1007/s12525-016-0219-0 Andam ZRB E-commerce and e-business/concepts and definitions. Wikibooks. https://en.wikiboo ks.org/wiki/E-Commerce_and_E-Business/Concepts_and_Definitions. Accessed 12 May 2016 Arnold D, Isermann H, Kuhn A, Tempelmeier H (2008) Handbuch logistik, 3rd edn. Springer Banker S, Cunnane C, Reiser C Logistics and supply chain trends to monitor in 2016. In: Logistics viewpoints, 11 January 2016. https://logisticsviewpoints.com/2016/01/11/logistics-and-supplychain-trends-to-monitor-in-2016/ Bhaskar W Top 6 technology trends in logistics and transportation industry for 2016. DreamOrbit, 22 January 2016. http://dreamorbit.com/top-6-technology-trends-in-logistics-and-transportation-in dustry-for-2016/ Brusch M, Eva S (2013) Trends in logistics in the German e-commerce and the particular relevance of managing product returns. LogForum 9(4). http://search.ebscohost.com/login.aspx?direct=tru e&db=edsdoj&AN=edsdoj.66fea15a4b874829833f6623392b99cd&lang=de&site=eds-live Buckley B E-commerce creating new niche for projects. ENR: Eng News-Record 277(14):41–41, 7 November 2016

26 Trends in E-commerce, Logistics and Supply Chain Management

609

Carr S Supply chain logistics trends to watch. Handshake Blog, 27 June 2016. https://www.hands hake.com/blog/supply-chain-logistics-2/ Cerasis (2015) Cerasis white paper: e-commerce & logistics Constine J, Escher A (2015) The mary meeker internet trends 2015 report. TechCrunch. http://soci al.techcrunch.com/2015/05/27/the-mary-meeker-internet-trends-2015-report/ Council of Economic Advisers (ed) 15 economic facts about millennials, October 2014 Doyle S Five trends impacting the retail supply chain in 2016. Checkpoint Systems, 6 April 2016. http://us.checkpointsystems.com/five-trends-impacting-the-retail-supply-chain-in-2016/ Economic Times Snapdeal launches ‘instant’ to speed up delivery. The Economic Times, 31 July 2015. http://economictimes.indiatimes.com/industry/services/retail/snapdeal-launches-inst ant-to-speed-up-delivery/articleshow/48299476.cms?intenttarget=no Federal Ministry of the Interior of Germany (2011) Demography report—federal government report on the demographic situation and future development of Germany. https://www.demograf ie-portal.de/SharedDocs/Downloads/EN/Publications/Demography_Report_Germany.pdf?__bl ob=publicationFile&v=2 Gibson BJ, Clifford Defee C, Randall WS (2016) The state of retail supply chain—essential findings of the sixth annual study. http://harbertcenters.com/supplychain/files/6th_Annual_SRSC_Repor t.pdf Gillman D (2014) 4 ways predictive analytics is transforming manufacturing. ERP-Desk, 2014. http://it.toolbox.com/blogs/inside-erp/4-ways-predictive-analytics-is-transforming-manuf acturing-60819 Grafes JB (2013) Maximizing productivity in e-commerce warehousing and distribution operations. Inbound Logistics. http://www.inboundlogistics.com/cms/article/maximizing-productivity-in-ecommerce-warehousing-and-distribution-operations/ Harper L (2015) 5 bold predictions from oracle CEO Mark Hurd. Oracle Marketing Cloud. https:// blogs.oracle.com/marketingcloud/5-bold-predictions-from-oracle-ceo-mark-hurd Heutger M, Kückelhaus M (eds) (2016) DHL logistics trend radar 2016. DHL Customer Solutions & Innovation International Federation of Robotics (2016) Executive Summary world robotics 2016 industrial robots. IFR International Federation of Robotics Kaplan M B2B ecommerce growing; becoming more like B2C, 5 December 2016. http://www.pra cticalecommerce.com/articles/85970-B2B-Ecommerce-Growing-Becoming-More-Like-B2C Khurana A 3PL: ecommerce logistics, warehousing, fulfillment. The Balance. https://www.thebal ance.com/3pl-ecommerce-logistics-1141740. Accessed 6 December 2016 Lamont J (2016) E-commerce: managing complexity. In: KMWorld. edsgao Laudon KC, Laudon JP (2014) Management information systems: managing the digital firm, Thirteenth edn. Prentice Hall, Boston Laudon KC, Carol GT (2015) E-commerce: business, technology, society. Pearson Longman N (2016) Top 10 logistics trends for 2016. Supply Chain Digital. http://www.supplychai ndigital.com/top10/4289/Top-10-Logistics-Trends-for-2016 Mangiaracina R, Marchet G, Perotti S, Tumino A (2015) A review of the environmental implications of B2C e-commerce: a logistics perspective. Int J Phys Distrib Logist Manag 45(6):565 Meeker M (2015). Internet trends 2015—code conference, KPCB (slide 171). See https://www.sli deshare.net/kleinerperkins/internet-trends-v1/171-171India_ECommerce_Leaders_More_Mobi lized McAfee A, Erik B, Davenport TH, Patil DJ, Barton D (2012) Big data: the management revolution. Harvard Bus Rev 90(10):61–67 McKinnon A, Sharon C, Anthony W, Michael B (2010) Green logistics: improving the environmental sustainability of logistics. Kogan Page Publishers Miller G Why harnessing ‘big data’ is the next big thing in ecommerce—part 1. UpstreamCommerce, 9 April 2012. http://upstreamcommerce.com/blog/2012/04/09/harnessing-big-data-big-ecomme rce-part-1

610

G. Sandhaus

Mollenkopf D, Stolze H, Tate WL, Ueltschy M (2010) Green, lean, and global supply chains. Int J Phys Distrib Logist Manag 40(1/2):14 Morganti E, Saskia S, Corinne B, Laetitia D, Barbara L (2014) The impact of e-commerce on final deliveries: alternative parcel delivery services in France and Germany. Transp Res Proc 4:178–90 Realisable Integrating WMSs or 3PL with your ERP. Realisable, 25 February 2013. http://www.re alisable.co.uk/integrating-wmss-or-3pl-with-erp/ Singh A, Aditi P (2016) E-commerce in India—consumer buying behaviour and growth trends an exploratory study. Asian J Multidiscip Studies 4(2). http://www.ajms.co.in/sites/ajms2015/inde x.php/ajms/article/view/1653 Sinha A, Paul W eCommerce supply chain insights in groceries and consumer packaged goods in the United States. Technology and Operations Department, Ross School of Business, University of Michigan, Ann Arbor MI 48109, 11 March 2015. http://www.tauber.umich.edu/research/%5Bwhite-papers%5D/ecommerce-supply-chain-i nsights-groceries-and-consumer-packaged-goods-united Turban E, King D (2012a) Dell—using e-commerce for success. Pearson. http://wps.pearsoned.co. uk/wps/media/objects/13897/14231103/additional_online/Online_Files_Ch01.pdf Turban E, King D (2012b) Electronic commerce: a managerial and social networks perspective. Global Edition Ulanoff L (2014) Amazon knows what you want before you buy it. Mashable. http://mashable.co m/2014/01/21/amazon-anticipatory-shipping-patent/ USA Today Beyond jobs numbers, labor shortage looms: our view. USA Today, 7 May 2015. http://www.usatoday.com/story/opinion/2015/05/07/jobs-numbers-labor-shortage-baby-b oomers-inflation-editorials-debates/70973518/ Wessel R, Greenway T (2014) Insight on: e-commerce & collaboration. Deutsche Post

Chapter 27

Multi-agent Systems Martijn Mes and Berry Gerrits

Abstract Within a Multi-Agent System (MAS), multiple autonomous agents negotiate, cooperate, and perform actions based on the goals and preferences of the (realworld) entities they represent. This chapter provides an introduction into MASs. In Sect. 27.1, on the basic level, we introduce the concept of agents and MASs, address their history, formally define MASs, and provide an overview of associated design methodologies. We illustrate one of these design methodologies using a case study of the use of Automated Guided Vehicles (AGVs) for trailer docking at a distribution centre. Next, we specifically focus on the use of MASs for manufacturing and logistics. First, in Sect. 27.2, on the advanced level, we discuss MASs for controlling the logistics processes within a company. We provide the motivation for using MASs in these environments, briefly review the literature on this topic, and present a case study on MAS control of AGVs in an industrial bakery. Next, in Sect. 27.3, on the state-of-the-art level, we focus on MASs to align the processes between different companies. We provide a brief overview of existing applications, describe the main challenges involved in inter-company MASs, and present a case study on MASs in the Port of Rotterdam.

27.1 Multi-agent Systems and Their Design (Basic) The agents considered throughout this chapter are proactive pieces of software that make decisions based on the preferences of their principals. Before formally defining agents, we first illustrate the concept of agents and how the behaviour of individual agents in a system with multiple agents influences the behaviour of the system as a whole. For this purpose, we use an example of an agent system from nature: the M. Mes (B) · B. Gerrits Department of Industrial Engineering and Business Information Systems, University of Twente, Enschede, Overijssel, The Netherlands e-mail: [email protected] B. Gerrits e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_27

611

612

M. Mes and B. Gerrits

Fig. 27.1 Ants building moving bridges (Chris Reid)

ant colony. Studies on ant colonies show that complex structures can be created at the colony level while the individual ants have no overview on how their work influences the colony as a whole (Franks 2009). A single ant has no powerful capabilities, however, when putting thousands or even millions of them together, the group collectively shows behaviour that far exceeds the capabilities of all the single ants put together (see Fig. 27.1). This so-called emerging or super additive property of such systems can be defined as that the performance of the parts put together exceeds the sum of the performances of the individual parts. Some have also called this a superorganism with collective intelligence. The intriguing beauty of such systems is that the collective intelligence emerges from the behaviour of all the individual ants without any centralized control. There is no supervisory ant to guide the ants to places where they can find food or to divide the tasks of building an ant hill, yet their collective behaviour enables the ants to find food and build large ant hills with complex tunnelling and temperature controlled chambers. The field of multi-agent systems (MASs) studies and applies the concepts of such natural phenomena to other real-world problems where centralized control might not be suitable (e.g., in dynamic and complex environments) or preferable (e.g., a central unit might not reflect the interests and preferences of the individual agents adequately). The remainder of this section addresses the history of MASs (Sect. 27.1.1), provides a definition of MASs (Sect. 27.1.2), presents various MAS design methodologies (Sect. 27.1.3), and introduces a case study (Sect. 27.1.4) used to illustrate one of the design methodologies (Sect. 27.1.5).

27.1.1 History Literature on MASs has been around since the mid-1990s and has gained popularity in recent years in the scientific community as well as in industry. The comprehensive book of Wooldridge (2009) on MASs mentions the following five important trends in the history of computing that contribute to the acclaim of MASs.

27 Multi-agent Systems

613

• Ubiquity is the constant increase of computing power against lower costs, which has led to increased processing power in many applications and devices. This omnipresence of technology has transitioned our society and businesses from closed and ill-informed to open and well-informed. • The interconnection of technology has led to large, complex distributed systems, which are typically more powerful than stand-alone systems. For example, smartphones enable us to share and process information at anytime, anywhere with anyone across the globe. But also businesses delegate their processing power across the globe to solve large-scale problems. • Increasing intelligence of systems also provides new opportunities for people and businesses alike. For example, navigation systems are able to pro-actively alter planned routes to find faster routes to the user’s destination based on current traffic information or road closures. • Another important trend is the continuing delegation of control to software systems. Nowadays we trust software to outperform humans in certain safety-critical tasks. For example, consider an automatic pilot on an aircraft or the advanced electronics in passenger cars to assist the driver or even take control of driving (e.g., adaptive cruise-control). • A final trend is the human-oriented view of programming and software. Increasingly, software systems are user-oriented with user-friendly Graphical User Interfaces (GUIs) and real-time interaction with the user. For example, consider Google’s “OK Google” or Apple’s “Siri”, which processes voice input and displays information based on the user’s input. These five trends pose challenges to software developers as systems become more intelligent, interconnected, and need to act on behalf of the user. To resolve these problems, the field of MASs emerged. A MAS consists of multiple agents with individual goals. To achieve these goals, they need to interact with each other, typically using communication protocols or message passing. In order to interact, agents must be able to cooperate, coordinate and negotiate with each other, similarly to how humans behave in their everyday lives (Wooldridge 2009). It may come as no surprise that this field has many related fields and builds upon issues from multiple disciplines. The main related disciplines are (i) artificial intelligence, (ii) humancomputer interface design and (iii) agent-oriented programming. The latter can be viewed as replacing objects with agents, as in real-world systems agents can better capture the essence of the individual entities compared to objects.

27.1.2 Definition, Characteristics and Architectures The concept of MASs is relatively easy to grasp. The main idea is that an agent is a computer system that acts autonomously on behalf of its user to achieve its design objectives. This means an agent can determine its own course of action instead of following a pre-determined code. Although there are many different definitions

614

M. Mes and B. Gerrits

of agents (Parunak 1998), the most commonly adapted definition throughout the literature is the following: An agent is a computer system that is situated in some environment and that is capable of autonomous action in this environment in order to meet its design objectives. (Wooldridge 2009)

From this definition, the following characteristics can be derived (Padgham and Winikoff 2005): • Situated: Agents are situated in an environment, which is most often dynamic, unpredictable and unreliable. • Autonomous: Agents can work independently of other agents as well as without interference of human intelligence. • Social: Agents affect other agents or may be affected by other agents. Communication and interaction is key in MASs. • Flexible: Agents should consider different environmental states and configurations and act properly to different situations to achieve their goals. • Proactive: Agents should continuously and persistently pursue their goal(s) and take the initiative when necessary. • Robust: Agents must be able to overcome failures by continuing to achieve their goal despite of previous failed attempts. Another important aspect to further refine our understanding of the concept of MASs is the control architecture. The control architecture describes how the agents are interrelated, as well as the hierarchy of agents. In general, the literature on control architectures distinguishes between four types: (i) centralized, (ii) proper hierarchical, (iii) modified hierarchical, and (iv) heterarchical. These four control architectures are depicted in Fig. 27.2. We briefly introduce these four architectures together with their advantages and disadvantages. The interested reader is referred to Dilts et al. (1991) for a more elaborate discussion. • Centralized control. In a centralized control architecture, all the decision making capabilities are within one central entity. The central entity sends commands to the lower level (non-intelligent) entities. All responsibilities are thus located at one central location and the lower level entities (which might be dispersed in a physical

Centralized

Proper hierarchical

Modified hierarchical

Heterarchical

Fig. 27.2 Control architectures (square  decision making entity, circle  execution only)

27 Multi-agent Systems

615

environment) have no say in what to do or how to do it. An obvious advantage of centralized systems is that there is complete information of the system. This enables the central controller to optimize the system (e.g., production planning). However, communication is a major disadvantage of a centralized system as all (local) information has to be send to the central controller at a sufficiently high frequency and with minimal delay. • Proper hierarchical control. Opposed to the centralized control architecture, this form has multiple control units and a master-slave relation between the hierarchical levels. The control units can be seen as supervisors of (a set of) subordinate entities and use aggregate information from lower levels for decision making. The control units delegate rather than prescribe jobs to the lower level entities. At the lower level, detailed decisions are made on how to perform the delegated action. For example, a control unit might send a set of jobs to a machine supervisor and the supervisor may determine on which machines and in which sequence the jobs are processed. This form of control does not suffer from scalability issues as adding more control units does not influence the (sub-hierarchy) of the other control units. On the other hand, problems may arise when the master control unit does not have sufficiently up-to-date or detailed information and the slave unit does not have sufficient capabilities to make decisions on the delegated job. • Modified hierarchical control. The main difference between proper and modified hierarchical control is the degree of cooperation between the subordinates. This loosens the master-slave relationship as subordinates can communicate (e.g., a supervisor might ask another supervisor to process some of its jobs). This proves to be more flexible and robust compared to the proper form. However, by adding horizontal collaboration, the complexity of the system increases. • Heterarchical control. This control form is the exact opposite of centralized control, since all of the decision capabilities are placed at the lowest level of the hierarchy. It is characterized by fully distributed local autonomous control units. The concept of global or aggregate information is absent in this control architecture. From the definition of MASs, one might deduce that MASs have a heterarchical control architecture by definition. This is by no means the case as autonomous and proactive decisions can be made at multiple levels in the control hierarchy. As an example, consider a road intersection with traffic lights, where stop-and-go decisions need to be made for the vehicles approaching the intersection. Obviously this decision is not made on the decentralized level (by the vehicles themselves) and also not on the centralized level (e.g., by the municipality’s traffic management system). Instead, an intermediate level is preferred, namely traffic lights that control a single intersection within the entire traffic system. However, nearby traffic lights might also collaborate to create so-called green waves to improve traffic flow. This horizontal collaboration resembles elements of the modified hierarchical control. Another option is, given the current advancements in sensor technology in cars, to remove the traffic light and make stop-and-go decisions based on cooperative behaviour among cars. This is an example of fully decentralized control.

616

M. Mes and B. Gerrits

27.1.3 Design Methodologies Building MASs is a complex and iterative process. Large-scale systems may contain hundred or more agents operating at real-runtime, perceiving, communicating, negotiating, and making decisions. Such a complex system has to be designed properly, by including all functionalities necessary to keep the system running effectively and efficiently, but also by making the system scalable and to some extent generic. The key challenge is to design a system such that it can be configured (e.g., by adjusting parameters or introducing new instances of generic building blocks) to represent a new situation and thus not to design a one-off system, which is only applicable to a certain architecture or testbed. On the other hand, we should avoid the pitfall of designing the system overly generic. A proper design methodology is required to balance these challenging requirements. Three well-known and widely used design methodologies are: Prometheus, Gaia, and MASE. Roughly speaking, these three methodologies consists of the following steps: (i) decomposition of the system into functionalities, (ii) allocation of the functionalities to agents, (iii) establishing interaction protocols between the agents, and (iv) designing the decision making capabilities of the agents. The first step is usually achieved by listing all system goals and grouping related goals. These related goals, together with related data, triggers, and actions, form functionalities. The main task here for the system designer is to decide among alternative decompositions. In the second step it is decided how these functionalities are allocated to the agents. In the third step, we face several design choices such as the sequence of steps in an interaction protocol. In the last step, we have to design protocols and algorithms for the inner working of the agents. This involves the way they react to triggers and incoming messages. The first three steps together form the architectural design phase and the last step is denoted by the detailed design phase. The architectural design phase is generally supported by Agent Oriented Methodologies. An important element within the second step, i.e., the allocation of functionalities to agents, is to decompose the system into multiple agents. There are two options for this: a functional decomposition, where agents represent functionalities or activities, and a physical decomposition, where agents represent one or more physical entities. In a manufacturing setting (see Sect. 27.2.1 for more information), the functionalities include order processing, product design, engineering analysis, process planning, production planning and scheduling, simulation, and execution. When applying a physical decomposition approach in manufacturing, the agents typically represent manufacturing resources (e.g., machines, robots, tools, fixtures, AGVs, and operators), aggregations of resources (e.g., cells, production lines, and shop floors) as well as products, parts, and operations. To illustrate the typical steps within a MAS design methodology, and to show the benefits of utilizing a structured design process, we consider two case studies. First, in Sect. 27.1.4, we apply the Prometheus methodology to a case study at a distribution centre. Later, in Sect. 27.2.2, we zoom in on some of the design choices in the architectural design phase using a case study at an industrial bakery. Both cases con-

27 Multi-agent Systems

617

sider the use of Automated Guided Vehicles (AGVs). AGVs are increasingly being used in logistics and industrial environments due to their flexibility, modularity, and increased productivity; see Vis (2006) for a comprehensive survey on AGV systems. Traditionally, AGVs utilize a central controller to assign the transportation tasks. However, for reasons mentioned in the beginning of this chapter, there is an increasing interest in utilizing decentralized control. Liu et al. (2002) provide a comparison between centralized and decentralized control, and conclude that their decentralized system provides higher utilization and is more robust to fluctuations in processing times. In a similar fashion, Mes et al. (2007) compare the performance of a MAS with two hierarchical control approaches. Using a case study of an underground AGV system around Amsterdam Airport Schiphol, they found that a properly designed MAS outperforms the other approaches in terms of vehicle utilization and service level.

27.1.4 Case Study: Autonomous Vehicles at a Distribution Centre In this case study we have a rectangular shaped distribution centre (DC) with multiple docking gates at both sides of the DC as shown in Fig. 27.3. The DC functions as a cross-dock facility, which unloads goods from inbound semitrailers and distributes these goods among outbound semi-trailers. The DC has a large terrain where yard tractors (YTs) transport the semi-trailers to and from the docking gates. We refer to this terrain as the apron space. The company is interested in replacing the manually operated YTs by AGVs. At the DC, truck drivers arrive with their semi-trailers, leave their semi-trailer at a parking area, and leave the DC again (possibly after picking up another outbound semi-trailer). The handling processes at the DC are carried out by the AGVs, which are able to pick-up a semi-trailer, drive it around the apron space, and rearward manoeuvre the semi-trailer into a docking gate. This decreases the waiting time of the truck driver at the DC and increases the handling per-

Entrance

(Un)loading docks

Arriving/departing trailers

AGV parking

Empty trailer parking

Fig. 27.3 Layout of the apron space of a DC

618

M. Mes and B. Gerrits

formance on the apron as automated vehicles generally outperform human operators and are more robust and reliable. A major technical requirement for such an automated DC is a vehicle control system that is able to (i) rearward an AGV with a semi-trailer and (ii) handle different types of semi-trailers (different dimensions, varying number of axles, and uneven load distributions of the cargo in the semi-trailer). Furthermore, we require a sophisticated planning and control system on the logistics level such that the AGVs handle the cargo in an efficient way. Within the planning and control system, decisions on, for example, vehicle dispatching, routing, conflict resolution, and battery management have to be made. Due to the stochastic and dynamic nature of transportation (e.g., uncertain arrival times due to traffic congestion), the planning and control system should also be able to cope with these dynamics. Hence, we propose using a MAS for the planning and control of the AGVs. We formally define this system as a closed transportation network consisting of a fixed number of pick-up and drop-off locations. AGVs transport the semitrailers between these locations using a certain track layout. The network is closed as no AGV can enter or leave the system, even when idling. Similarly to Ebben (2001), this Automated Transportation Network is defined as a fully automated system for the transportation, loading, and unloading of goods using AGVs, supported by a MAS that functions as a planning and control system. In order to systematically address the challenges in this case study, a comprehensive and detailed methodology should be used to guide the process of breaking down the high-level objective ‘building an intelligent multi-agent system’ into smaller, easier to grasp chunks, such that the system can be fully understood and then designed. This methodology should not only include highlevel steps such as ‘specifying the system boundaries’, but also mid- and lower level steps providing enough detailed guidelines to design and implement intelligent agents.

27.1.5 Application of the Prometheus Design Methodology We illustrate the process of designing a MAS using our case study. As mentioned before, we here consider the Prometheus methodology. This methodology is specifically designed to include agent-technology and aims to be complete, providing everything necessary for specifying and designing agent systems (Padgham and Winikoff 2005). Furthermore, it comes with the freely available Prometheus Design Tool (PDT), which has built-in completeness and consistency checks. We will not provide an elaborate step-by-step discussion on the Prometheus Methodology, for which we refer to Padgham and Winikoff (2005). Instead, we highlight and summarize the key steps of the methodology tailored to our case study.

27 Multi-agent Systems

619

The Prometheus methodology consists of the following three phases: (i) the system specification phase that focuses on identifying the goals and basic functionalities of the system, along with inputs (percepts) and outputs (actions), (ii) the architectural design phase that uses the results from the previous phase to determine which agent types the system will contain and how they will interact, and (iii) the detailed design phase that looks at the internals of each agent and how it will accomplish its task within the overall system. In the following paragraphs, we summarize the results of these design phases for our case study. For more details on the MAS design related to this case study, we refer to Gerrits et al. (2016). System Specification In the system specification phase, we break down the overall system goal into smaller sub-goals to elucidate which kind of behaviour and functionality is required to reach the system goal. We define the system goal as follows: To design a generic automated planning and control system based on agent technology for the pick-up and delivery of semi-trailers by means of AGVs in a collision- and conflict free environment in such a way that it yields a cost-effective, near-optimal solution. From this initial description we first deduce (and later refine) a list of system goals. To accomplish these goals, several sub-goals can be defined, such as path planning, vehicle dispatching, and congestion avoidance. From this breakdown, we create a network of connected goals and group these goals using the standard software engineering criteria of coupling and cohesion to form functionalities. Coupling is the level of interdependency between functionalities, while cohesion is the level of uniformity of the goals in a functionality. From this analysis we define the following (generic) grouping of functionalities: • Demand Management (DM). This functionality monitors inbound/outbound semitrailers, obtaining information about expected arrival/dispatch times and cargo descriptions. • Park Management (PM). This functionality assigns pick-up and drop-off locations to all semi-trailers and AGVs. • Vehicle Scheduling (VS). This functionality decides when and where which AGV should pick-up and drop-off (maybe empty) semi-trailers. • Vehicle Routing (VR). This functionality determines the route such that the AGV can pick-up and drop-off (maybe empty) semi-trailers. • Conflict Resolution (CR). This functionality monitors all AGV movements, makes sure there are no collisions, and resolves conflicts. • Battery Management (BM). This functionality determines when and where AGVs should be recharged. The system specification phase continues with scenario development. Every scenario sets out a state the system can be in and translates this into how the functionalities defined above should act upon this state. We define the following scenarios: (i) cargo arrival, (ii) cargo departure, (iii) empty semi-trailer movement, (iv) AGV idle, (v) AGV low battery, and (vi) AGV conflict. The first scenario concerns inbound cargo and the second one outbound cargo. The third scenario occurs when an empty

620

M. Mes and B. Gerrits

semi-trailer is moved to (or removed from) a dock. The fourth scenario occurs when an AGV is idling and thus has to be parked or strategically positioned. The fifth scenario occurs when an AGV has a low battery and thus has to be recharged at an appropriate location. The final scenario is one that can happen anytime during system run-time and thus also during other scenarios, namely AGVs that are in conflict. This may happen, for example, when two or more AGVs want to take the same route at the same time. Appropriate measures have to be taken when this occurs such that congestion and system dead- or livelocks are avoided. Architectural Design The architectural design phase defines the agents based on the functionalities and scenarios defined in the previous design phase. In this phase, we determine the interrelationship between agents and which external connections are required. We employ a so-called data coupling diagram to provide insight into the connections between functionalities based on the defined scenarios, as shown in Fig. 27.4. Directed links are shown between functionalities and data. Arrows pointing towards data indicate that data is written by the functionality and reverse arrows indicate the usage of data. Double-headed arrows indicate both the usage and the writing of data by the functionality. Bold arrows indicate important connections within the system. All data used and produced by the system is placed in the diagram, extended to include all external data sources. The main external database required for the MAS

Board computer

AGV sensors

Planning system

7

6

TMS

AGV Manager

Demand Manager

AGV Status

1 Cargo DB

5

2

3

Parking Manager

Cargo Parking

Conflict Manager

AGV Parking

System boundary

Fig. 27.4 Architectural design

BaƩery Manager

4 Vehicle Scheduling

Vehicle RouƟng

Vehicle Schedules

Vehicle Routes

27 Multi-agent Systems

621

is the Transport Management System (TMS). This TMS contains information about inbound and outbound cargo, such as the destination at the DC (originating from the planning system) and client-specific information. The expected arrival time of inbound cargo is obtained using the board computer of incoming trucks. All links in the diagram originate from one or more scenarios as defined in the system specification phase. This diagram thus also functions as a consistency check with the scenarios defined in the first design phase. From this diagram we see seven agents emerging, similarly to the previously defined functionalities but extended with an AGV manager. For the sake of brevity we do not list the exact communication between the functionalities in the data coupling diagram as the communication is typically scenario-dependent. Detailed Design The final design phase builds upon the agents defined in the previous phase and further specifies the behaviour of each agent. This includes all triggers an agent responds to, how it responds, and with which other agents it interacts. This phase also narrowly defines the external connections of the MAS as these contain many of the triggers the MAS responds to (e.g., arrival of loaded semi-trailers or finished unloading a semi-trailer). The detailed design of every agent is captured in two diagrams: (i) the agent overview diagram and (ii) the capability overview diagram. The first one provides a top-level overview of the agent and the latter takes a more detailed view of all the capabilities of an agent. For more information on these diagrams we refer to Padgham and Winikoff (2005).

27.2 Multi-agent Systems for Production Processes and Internal Logistics (Advanced) In this section, we specifically focus on manufacturing environments, including internal logistics (e.g., material flow and warehousing operations). For these environments, it is difficult to use a single central controller due to complexities and uncertainties in data management, resource availability, and time lags between the physical and informational processes. The solution is to distribute control over multiple decisional entities. Since the early 1970s, distributed control was mainly organized hierarchically and based on the Computer Integrated Manufacturing (CIM) paradigm (Trentesaux 2009). Here, control is divided into hierarchically dependent sub-problems, where typically decisions are made in a top down manner with long-term (tactical) decisions made on the higher levels and short-term (operational) decisions on the lower levels close to the manufacturing resources. Typically, the exchange of information between units at the same decision level is not permitted in such (proper) hierarchical control systems (see Sect. 27.1). Despite the fact that hierarchical control systems may be efficient in environments with minor dynamic nature, they are less suitable for highly dynamic and complex environments. Global competition, rapidly changing market demands, shorter prod-

622

M. Mes and B. Gerrits

uct life cycles, the need for a more diverse product mix, increased customization in manufacturing, rapid advances in technology, and high uncertainties in the production processes, are forcing major changes in production and in the configuration of manufacturing organizations. To deal with the above mentioned trends, today’s production systems must be able to quickly adapt to disturbances while maintaining shorter product cycles, improving productivity, and increasing operational flexibility. Therefore, more flexible manufacturing concepts have been introduced that are reconfigurable and consist of autonomous and intelligent modules, which dynamically interact with each other to achieve local and global objectives (Caridi and Cavalieri 2004). A specific example is a Flexible Manufacturing System (FMS), which is a system that is capable of producing a variety of products with minimal time lost in changeover from one product to the next. According to Browne et al. (1984), two main types of flexibility can be included in the realization of an FMS: routing flexibility and resource flexibility. These flexibilities can be achieved using, e.g., multi-robot and AGV layouts as considered in the first two case studies in this chapter. The changing environment also has an impact on manufacturing planning and control. Control systems need to be flexible, reconfigurable, embed intelligence, and allow timely response. Time should be seriously considered as a limiting resource, particularly due to the shift from static to dynamic optimization in manufacturing environments. The centralized or hierarchical models are often not able to provide the required planning flexibility and timely response. Since the 1990s, the focus has been more on distributed control, since the required communication between the decision levels in a hierarchical system results in low response times and low robustness (Trentesaux 2009). In distributed control, the local decision making units do not necessarily need to ask permission for control actions to higher decision levels. Instead, the communication typically takes place between the local decision makers. In case there is no communication with higher level decision makers, we speak of heterarchical control systems, as introduced in Sect. 27.1. MASs have been widely used to model heterarchical control systems; this is the topic of Sect. 27.2.1. After that, in Sect. 27.2.2, we present a case study on MAS for heterarchical control of AGVs in an industrial bakery.

27.2.1 MAS to Support Manufacturing Planning and Control As argued before, MAS is a promising approach for modelling heterarchical control systems. By distributing the computational resources and capabilities among the agents, the system suffers less from the “critical point of failure” problem associated with centralized systems (Wooldridge 2009). Different from centralized approaches, agent-based manufacturing scheduling and control systems are able to quickly respond to dynamic changes and disturbances through local decision making. Besides the ability to model heterarchical systems, the key benefit of MAS is that the intelligent agents are able to learn and adapt to dynamic and changing environments.

27 Multi-agent Systems

623

One of the earliest MAS for manufacturing, called YAMS (Yet Another Manufacturing System), was developed by Parunak (1987). He considers a hierarchical production system in which every node in the control hierarchy (factory, manufacturing cell, workstation, and machine) is represented by an agent. Each agent has a collection of plans and negotiates with lower level agents to assign production tasks. In a later paper, Parunak et al. (2001) presented AARIA, where the agents represent manufacturing entities such as parts (e.g., materials and products), resources (e.g., machines and AGVs), and unit processes (specific tasks). This physical decomposition approach (see Sect. 27.1.3) to represent manufacturing resources and tasks by agents is common in agent-based manufacturing systems. A typical MAS structure for manufacturing planning and control contains the following agent types: resource agents, job agents, and product agents (Van Brussel et al. 1998). The resource agents represent production resources in the manufacturing system, and they contain knowledge about methods to allocate and control their corresponding resources. The job agents represent the manufacturing and logistical tasks, they hold all relevant task related data, and they are responsible for performing the corresponding work correctly and on time. The product agents hold all information related to the required production processes and components to produce the corresponding products. Typically, job agents focus on on-time delivery against the lowest possible costs, and resource agents strive for utilization and/or profit maximization. A key issue is how to configure the agents such that their self-interested behaviour yields a near-optimal solution for the network as a whole. To contribute to system-wide objectives, the agents need to interact with each other. According to Jennings et al. (2001), these interactions may involve (i) simple information exchanges, (ii) requests for particular actions to be performed, (iii) cooperation (working together to achieve a common objective), (iv) coordination (arranging related activities to be performed in a coherent manner), and above all (v) negotiation (the process of reaching a mutually acceptable agreement between several agents on a set of tasks). Agents need to interact with other agents to achieve their objectives either because they do not have sufficient capabilities or resources to solve a problem on their own, or because there are interdependencies between the agents that follow from being situated in a common environment (Faratin et al. 1998). Given a physical decomposition approach, typically job agents negotiate with resource agents to support the allocation and scheduling of tasks. For multi-agent interaction and negotiation, there are several applicable economic models. A frequently used negotiation form is the so-called Contract Net Protocol (CNP) from (Smith 1980), which is a high level protocol for achieving efficient cooperation in a similar way as companies tender contracts in public markets. It is the most common mechanism for distributed task allocation in agent-based systems (Wooldridge 2009). CNP consists of four steps: (i) an initiator sends a call for proposal to a set of participants, (ii) the participants respond with a proposal, (iii) the initiator chooses the best proposal and awards a contract to the respective participant, and (iv) the other participants are rejected. Other negotiation protocols include voting (all agents provide their preferences, the outcome is based on these preferences and imposed to all agents), and market-based protocols such as bargaining (any process

624

M. Mes and B. Gerrits

through which players reach an agreement) and auctions (an auctioneer that wants to sell an item and bidders that bid on that item). Regarding the latter, there are many different auction types. For one-to-many environments, we distinguish between the Dutch auction (descending: the seller continuously lowers the price until one of the bidders acquires the item at the current price), the English auction (first-price open-bid: all bids are public and the highest bidder wins), the first-price sealed-bid auction (bids are only known to the auctioneer), and the Vickrey auction (second-price sealed-bid auction: the highest bidder wins, but pays the price of the second-highest bid). For many-to-many auctions, double auctions are being used (single shot or continuously open for new items and bids). All auction types may involve a single item or multiple items. When bids on combinations of items are allowed, we speak of combinatorial auctions. For an extensive literature survey on auctions we refer to Mcafee and McMillan (1987). As mentioned above, the typical MAS for manufacturing planning and control consists of agents representing individual entities (resources, jobs, and products). In these heterarchical control structures, each agent individually finds the suitable agents (by using the negotiation mechanisms as discussed above) that are able to carry out its task without any degree of centralization. An alternative is to use mediation architectures. These architectures involve special agents, mediators or traders, to manage the allocation of tasks to agents (similar to the traffic light example in Sect. 27.1.2). The mediator agent should have full knowledge on the availability and capabilities of all involved agents. The advantage of this architecture is that the mediator possesses more knowledge, which might improve the system-wide performance (as opposed to the performance reached via individual agents’ decisions); however, it also reintroduces some of the disadvantages of hierarchical control as discussed before. For more information on mediation architectures we refer to Maturana and Norrie (1997) and Maturana et al. (1999). Within the manufacturing domain, MASs have received a growing interest following the trends and developments that are mentioned in the introduction of Sect. 27.2. However, industrial applications are still rare as argued in several survey papers. In an overview paper on MASs for intelligent manufacturing, Shen et al. (2006) state that one of the reasons for this lack of applications is the difficulty of full integration of manufacturing process planning, scheduling, and control, and integration of MASs with the existing information systems. Monostori et al. (2006) reach a similar conclusion, but also state that agent technology is attractive in all domains of manufacturing. Also Pˇechouˇcek and Maˇrík (2008) conclude that the potential of agent technology has not been fully utilized and provide an overview of potential obstacles for wider deployment. Reasons for the weak adoption in combination with research opportunities are also listed by Leitão (2009). In a later paper, Leitão and Vrba (2011) again state that adoption of MASs in the manufacturing field is rare, but that it has already widely been used in other domains, such as e-commerce, business applications, and logistics. In a recent survey of deployed examples of agent technology in manufacturing, Müller and Fischer (2014) come to a more positive conclusion. After analysing 152 applications, they conclude that agent technology has been suc-

27 Multi-agent Systems

625

cessfully deployed in a significant number of applications, and continues to be an increasingly useful and impacting technology in various sections. In the next section, we present a case study on MAS control of AGVs in an industrial bakery, utilizing the typical agent setup as presented in this section.

27.2.2 Case Study: AGV Control in an Industrial Bakery This case study involves the design and development of a MAS for an industrial bakery. The bakery, called Merba, is located in The Netherlands and produces a wide range of cookies for the international market. In this section, we briefly present the case and illustrate the MAS design choices. A more detailed description of this system can be found in Mes et al. (2008). The focus of this case study is on the dough preparation process. Currently, employees collect the ingredients for the dough manually in barrels and move these barrels between the various processing locations. This manual process has a negative effect on labour conditions, product quality, and traceability of ingredients. Therefore, Merba aims at a fully automated dough production process using AGVs. The requirements for the new system closely follow the general requirements for today’s manufacturing systems as mentioned in the introduction of Sect. 27.2: a flexible production system that (i) can easily be adjusted to new product introductions or modifications in the bakery layout, and (ii) can react in real-time to process uncertainties like equipment failures and the arrival of rush jobs. For these reasons, the management prefers a MAS for the control of AGVs. We now describe the dough preparation process, with a focus on the AGV movements. An overview of the different transportation tasks is given in Fig. 27.5. The dough preparation process starts with a dough production request generated by the Manufacturing Execution System (MES). Each dough request is restricted by an earliest- and latest delivery time of the dough at the production line. First, we have to find a suitable barrel, i.e., one that has not previously been used for an incompatible dough type. Contaminated barrels need to be cleaned first at a special cleaning area. An AGV picks up the barrel and moves it towards a storage area consisting of several silos with different ingredients. To collect the ingredients, the AGV positions its barrel below these silos. Next, the AGV moves the barrel to a mixer. Similar to the barrel selection, contamination issues have to be taken into account in the mixer selection. After mixing the ingredients, the AGV transports the dough to a rising area where it has to wait. For each dough type, a minimum and best rising time is specified. Once the minimum rising time has passed, the barrel can be moved to the production line. At the production line, the barrel is emptied in a feeder that delivers the dough to a conveyer belt moving through the oven. Because a feeder can store more than one barrel of dough, dough can be delivered some time before it is

626

M. Mes and B. Gerrits

actually needed, given by the earliest delivery time. The empty barrel will stay on the AGV or will be dropped at a barrel storage area. The main decisions that need to be made are displayed in Fig. 27.6. The overall goal is to minimize both the deviation from the best rising time (product quality) and the tardiness with respect to the latest delivery time. To achieve this goal with the least number of AGVs, we have to schedule the tasks as efficiently as possible. The main design questions when decomposing the system into a MAS are shown in Fig. 27.7. Because of the decomposition, the overall goal has to be realized by individual agents with individual goals. The main difficulty here is that the goals of individual agents may deviate from the overall goal, and even may be conflicting due to limited resource capacities. An example of divergence in goals is that minimizing the costs of one dough delivery may have a negative effect on the future availability of AGVs, possibly resulting in late deliveries. To design the MAS, we use the Prometheus methodology as described in Sect. 27.1. We specifically focus on some of the challenges in the architectural design phase, as the general steps are already illustrated in the first case study. For illustrative purposes, we ignore barrel availability, cleaning, and recharging, and consider only two functionalities: dough preparation management and dough delivery management. These functionalities are concerned with the allocation and scheduling of respectively preparation jobs (barrel storage area/cleaning—silos—mixer—rising area) and delivery jobs (rising area—production line—barrel storage area/cleaning). A preparation job may be scheduled immediately after release of a dough request. A delivery job is released once the dough has been delivered to the rising area. In the architectural design phase, we need to (i) define the agents, (ii) assign functionalities to these agents, and (iii) define agent interaction patterns. We use a physical decomposition approach, where agents represent physical objects in the bakery. Given the reduced scope, we have the following objects: AGVs, lines, silos and mixers. We decided to use an AGV agent, a line agent, and a storage agent that represents all silos and mixers. To assign the functionalities to these agents, we look at data and triggers. Functionalities may be triggered by actions of physical objects within the factory, which then form candidates for these functionalities. Functionalities that share the same data source form candidates for allocation to the same agent because it requires less information exchange. We investigate the allocation of the functionalities either to the line agent or to the AGV agent. If we allocate them to the line agent, which we denote by line centric architecture, the line agent will search for an AGV based on its triggers (e.g., it receives a dough request from the MES, or it is informed that dough has been delivered at the rising area). If we allocate these functionalities to the AGV agent, which we denote by AGV centric architecture, then the

27 Multi-agent Systems

627

AGV agent will search for a job at all lines based on its triggers (e.g., when it becomes idle). After allocating the functionalities to agents, we need to specify how agents exchange information to perform their given functionalities. Again we can use Prometheus by building scenarios, interaction diagrams and protocols (see Sect. 27.1.5). The main difficulty is to establish suitable interaction sequences that describe (i) which agents are involved in the information exchange, (ii) who communicates with whom, and (iii) what the timing of communication is. As an example, suppose the dough preparation management functionality is allocated to the line agents. Then, whenever a line agent receives a dough request, it might contact all AGV agents to be informed about their availability, and ask the storage agent about the availability of the silos and mixers. Alternatively, the AGV agents themselves might contact the storage agent upon a request from the line agent. Instead of first contacting the AGV agents, the line agent might also first contact the storage agent. One can imagine that we might end up with a large number of possible interaction sequences, see Mes et al. (2008) for an overview. For illustrative purposes, we now focus on only two agent types: AGV agents and line agents. As a result, we only have two interaction sequences from which we need to choose. In the AGV centric architecture (Fig. 27.8), the job allocation process is triggered by an AGV that becomes idle. The AGV agent sends a request to all line agents to submit their job characteristics. After receiving the job characteristics, the AGV agent selects the most suitable job and informs the corresponding line agent about the expected delivery time of this job. It might occur that a line receives a dough request while all AGVs are idle; in that case an arbitrary idle AGV is triggered. In the line centric architecture (Fig. 27.9), the preparation jobs are triggered by a line agent that receives a dough request, and the delivery jobs are triggered when dough has been delivered to the rising area. The line agent then sends the job characteristics to all AGV agents. Each AGV agent selects the best time to start this job and calculates a price. The line agent simply selects the AGV agent with the lowest price and informs the winning AGV agent. The last step in the architectural design phase requires describing the communication protocols. As can be seen in Figs. 27.8 and 27.9, we used the CNP to support the dough preparation and dough delivery management functionalities.

628

M. Mes and B. Gerrits Mixers

Preparation job

Silos

Delivery job Battery recharge

4

3

Ovens Rising

1

2

2 1

Recharge

Barrels

3

Cleaning

Fig. 27.5 AGV movements in the industrial bakery Which AGV?

Which barrel? Barrels

AGVs

Ovens

Silo’s

Waiting Mixers

Cleaning

Recharge

Fig. 27.6 Decisions to be made

27.3 Multi-agent Systems Within Supply Chains (State-of-the-Art) In this section, we first discuss the applicability of MASs for supply chain management (Sect. 27.3.1), present a case on this topic (Sect. 27.3.2), and end with a description of the main challenges related to this area (Sect. 27.3.3).

27 Multi-agent Systems

629

Which agents? Who communicate? Who initiates?

AGVs

Barrels

Ovens

Silo’s

Waiting Mixers

Cleaning

Recharge

Fig. 27.7 MAS decomposition AGV Idle

Schedule jobs

1 1

Evaluate jobs

Line

Request for proposals Award & Delivery time Job characteristics

1

n 1

Dough request Create jobs

Dough delivery

n

Fig. 27.8 AGV centric architecture

Line Select lowest price

Dough delivery

1

Winner

1

Dough request

AGV

Price

Create jobs

n 1

Schedule jobs

Job characteristics 1

Price jobs

n

Fig. 27.9 Line centric architecture

27.3.1 Applicability of MAS for Supply Chain Management A supply chain consists of all companies involved in the production and distribution of products, from the acquisition of raw materials to their delivery at the end consumers. These companies include, e.g., suppliers, factories, warehouses, distribution centres, retailers, and transportation companies. For many years, the supply chain companies have focused primarily on their individual operations, and supply chain relations were based on long-term relationships among partners. However, due to increasing competition, reduced profit margins, increasing market globalization, the advancement of e-commerce, and continuous changes and uncertainties in the supply chain (changing consumer requests, resource unavailability’s, breakdowns, etc.),

630

M. Mes and B. Gerrits

this approach is no longer sufficient. Nowadays, companies are exploring new forms of organization and collaboration mechanisms to work together with their suppliers and customers. Supply Chain Management (SCM) is a strategy through which the different supply chain functions (marketing, distribution, planning, manufacturing, and purchasing) can be integrated. SCM involves the coordination of independent enterprises to improve the supply chain performance as a whole while satisfying individual companies’ needs as well. A typical example to illustrate the importance of supply chain management is the so-called Bullwhip effect, see Lee et al. (1997). The Bullwhip effect refers to the problem in distribution channels (factories, distribution centres, wholesales, and retailers) where the fluctuations in demand and hence final inventory levels are amplified upstream in the supply chain. These amplifications are the direct result of individual companies’ limited visibility of the demands from end customers. Indeed, the Bullwhip effect can be avoided by sharing actual market demands across the supply chain. Although the benefits of integrating supply chain functions are apparent, supply chain members have difficulty to find coordination mechanisms through which they can align their activities (Douma 2008). The coordination mechanisms should allow for synergies while guaranteeing the interest of each individual company. The reasons why collaboration is hard to achieve are various. First, each company in the supply chain has its own interests and goals. Centralized coordination mechanisms may find it hard to cope with these possibly conflicting interests. Second, companies may be reluctant to give up part of their autonomy, especially in situations with one strong leader in the supply chain (e.g., an OEM) that might dominate the others. Third, companies might be reluctant to share information with each other, especially competitive and privacy sensitive information. Fourth, supply chains are complex in the sense that companies may join or leave the chain, and most enterprises nowadays participate in a number of chains. Finally, centralized planning approaches might have difficulty with the dynamic nature of supply chains, i.e., they might be sensitive to information updates and might not be able to respond in real-time to information changes. MASs are able to cope with many of the challenges mentioned above. They provide a natural approach to model supply chain networks, where the supply chain is represented by a set of cooperating intelligent agents that each perform one or more supply chain functions. In a MAS for SCM, often a combination of a physical and functional decomposition is used. Physical elements typically represent the individual participants, such as retailers, distribution centres, plants, suppliers, and transportation companies. Functional elements correspond, e.g., with inventory, demand, supply, and transportation management. Typically, mediator agents are introduced to support cooperation between the supply chain partners, in combination with negotiation protocols to overcome conflicts and to reach agreements among agents. We distinguish four different viewpoints with regard to MAS for SCM, partly based on Ahn et al. (2003) and Um (2010). First, MAS as a modelling and simulation approach. Then, supply chains are modelled and analysed through agent-based simulation, where the performance of various supply chain designs can be evaluated.

27 Multi-agent Systems

631

Second, MAS as an information system that supports the alignment of operations of different companies through information exchange. Typically, the different supply chain companies use different software systems, which can be connected through a MAS. Third, MAS to support the dynamic formation of supply chains through multiagent negotiation. Agents perform negotiations with other agents to form a supply chain in a certain period to achieve a common objective. Finally, MAS to support the coordination of supply chain operations to improve operational performance. Here, the intelligence of agents often relies on operations research and machine learning techniques. Coordination mechanisms are used between the different supply chain partners to align the operations throughout the supply chain. Typically, a MAS implementation for SCM exhibits multiple viewpoints as mentioned before. For example, only electronic integration (second viewpoint) might not result in improved decision making. In the next section, we present an example case on MAS for SCM. After describing the case, we end with a list of challenges involved in MAS for SCM.

27.3.2 Aligning Barge and Terminal Operations in the Port of Rotterdam We address a complex supply chain coordination problem, namely the alignment of barge and terminal operations in the Port of Rotterdam. Barges visit the port to load and unload containers at multiple container terminals. For the handling of ships, appointments are made between terminal and barge operators, typically by telephone, fax, and e-mail. Unfortunately, it happens frequently that appointments are not met. Due to the fact that barges have to visit multiple terminals, a disruption at one terminal may propagate through the port and disturb the operations of many other barge and terminal operators. The result is that barge operators face long and uncertain waiting times at terminals, and that terminals face uncertain arrival times of barges resulting in idle times of their quay resources. This in turn leads to high direct costs, environmental pressure, and congestion on the road and rail infrastructure. Solving the problem improves the hinterland connectivity and thereby the attractiveness of the port of Rotterdam significantly. This alignment of barge and terminal operations is a clear example of an interorganizational coordination problem. For reasons mentioned in Sect. 27.3.1, central coordination of all activities in the port is not an acceptable option. Additional complicating factors are that terminal operators compete with each other (and so do barge operators) and that no contractual relationships between terminals and barges exist, which means that no performance can be contractually enforced. For these reasons, we propose to use a MAS. The design consists of only two agent types: Barge Operator Agents (BOAs) and Terminal Operator Agents (TOAs). All agents from the same type have identical intelligence. Below, we describe the resulting MAS design, focusing on the basic information exchange as shown in Fig. 27.10.

632

M. Mes and B. Gerrits Multi-agent system 1. Request rotation

2. Request STP from all terminals

Input berth plans (asynchronous) 3. Respond with STP

4. Rotations

5. Select rotation 6. Request confirmations

7. Confirm appointments Barge operators

Barge operators agents

Terminal operator agents

Terminal operator

Fig. 27.10 MAS to align barge and terminal operations

Barges arrive in the port over time. Upon arrival in the port, the barge operator requests information from the terminals it has to visit regarding their availability. The information requested by a BOA from a TOA is incorporated into a so-called Service-Time Profile (STP), which denotes a guaranteed latest departure time given an arrival time at the terminal, for all possible arrival times during a certain planning horizon. A terminal has to reply instantaneously with an STP having only limited knowledge about future arriving barges. Terminal operators can use the STPs to indicate preferred handling times thereby optimizing their capacity utilization. After receiving all STPs, the BOA calculates the best five rotations, where a rotation defines a route along all terminals to be visited. The barge operator planner chooses one of the proposed rotations and announces its preferred arrival time at the terminal. The terminal operator makes an appointment by confirming the barge’s latest arrival time and a guaranteed latest departure time. In case one or more terminals are not able to confirm the requested appointments (due to the fact that, in the meantime, appointments have been made with other barges), the complete rotation is cancelled, and the process starts all over again, until a rotation can be completely booked. When barges arrive after their latest arrival time, the appointment will be cancelled, regardless of the reason for the delay. During the entire process from planning to execution, the terminal has to deal with uncertainty and disturbances, such as uncertain arrival times and handling times of barges and container vessels, as well as cancellations and no-shows. The TOAs and BOAs are using a variety of algorithms to create STPs and barge rotations respectively. The idea behind using STPs is that terminals do not have to share competitive sensitive information. For example, a high service time (difference between the latest departure time and the latest arrival time) might indicate many visits at this terminal, but could also mean that there is no crew scheduled at this time. But still, it provides enough information for barges to plan their rotations efficiently. For more details on this case study, we refer to Douma et al. (2011).

27 Multi-agent Systems

633

Although the proposed MAS overcomes many of the challenges posed earlier, it is still a challenge to implement this system. One way to gain acceptance, is through the use of a management game. For this specific case, a game has been developed where each player takes the role of a barge operator. Within the game, four different rounds can be played, that differ in the amount of information exchanged and the amount of decision support given by the BOAs. By playing these four rounds, we were able to demonstrate that (i) more information on available terminal capacity results in better routes, (ii) the use of decision support from the BOA results in the best routes, and (iii) an overall increase in performance, compared to a round without information exchange and decision support, can be achieved without central coordination. For more details on the game, we refer to Mes et al. (2014).

27.3.3 Challenges Involved in Inter-company MAS The design of a MAS for SCM introduces many additional challenges compared to a typical MAS application within the context of one company. These challenges explain the relatively few reported successful implementations of MAS for SCM, and require more scientific research to cope with them. We distinguish between seven main challenges. The first challenge is related to regulation within MAS. To arrive at a stable system, it is necessary that agents are not able to manipulate each other, e.g., by providing incorrect information. To make manipulation or strategic behaviour unattractive, we need to regulate the behaviour of the agents. This can be achieved through external regulation or self-regulation. External regulation is realized by an external party that registers the behavior of individual participants and can penalize a participant if it displays undesired behavior. Self-regulation means that the participants in the system correct each other without the help of an external party. The second challenge is related to standardized communication and negotiation protocols, which is needed to integrate the different software systems of the different supply chain partners. The most widely used Agent Communication Languages (ACLs) are the ACL defined by the Foundation for Intelligent Physical Agents (FIPA) and the Knowledge Query and Manipulation Language (KQML). A negotiation protocol is used to organize message sequences among agents. Negotiation protocols specify (i) the way agents interact, (ii) the kind of deals that the agents can make, and (iii) the sequence of offers and counter-offers that are allowed. In Sect. 27.2 we already mentioned some typical protocols. The third challenge involves the freedom of the participants to design their own agents. For the case study example, we need to decide whether barge and terminal operators are allowed to design the full functionality of their agent themselves or only to configure their agent. The fourth challenge relates to the responsiveness of the system. In a dynamic environment, the time available to respond may vary based on the event. An agent must be able to respond within the time allotted. For our case study, if a barge operator

634

M. Mes and B. Gerrits

takes too much time to make a decision, the state of the system may have changed in the meantime, e.g., certain time slots at terminals might not be valid anymore because another operator agent applied for it. The fifth challenge is related to gain sharing. The sharing of benefits and losses is complex and strongly relates to the field of game theory. The gains and losses need to be distributed fairly and acceptably among all participants. For our case example, we can see that this is not easy. For example, given the interdependencies in the system, it is difficult to trace a delay back to a specific barge or terminal. The sixth challenge is concerned with the degree of autonomy of agents. If agents only provide decision support that has to be confirmed by a human (the principal of the agent), one can imagine that delays in the decision process are likely to happen. A fully autonomous agent, however, makes decisions for its owner. In this case, the principal needs to trust its agent and has to be convinced that the agent decides in its best interest and that no better performance can be achieved. The final challenge relates to the business and governance model. When supply chain partners agree on the implementation of a MAS, various questions arise, like who is going to implement and manage the system, who is going to pay for it, who will be responsible for the functioning of the MAS, who will take care that participants live up to the agreements made, and who will be the owner?

27.4 Further Reading Wooldridge (2009) provides a general introduction into MASs. More information on designing and building MASs and the Prometheus design methodology can be found in Padgham and Winikoff (2005), while a general introduction to complexity theory and complex systems is given by Mitchel (2009). An overview of research and developments of industrial applications of MASs is presented by Leitão et al. (2013), including a review of methodologies, architectures, applications, trends, challenges, and future application domains of industrial agents. Tools necessary for implementing a flexible, robust, adaptive, and fault-tolerant manufacturing system are discussed by Paolucci and Sacile (2016). With respect to supply chain planning, Barbati et al. (2012) review various optimization problems, while collaborative supply chain planning by means of agentbased models are discussed by Hernández et al. (2014). Chaib-draa and Müller (2006) finally present an overview of developments, literature, and future challenges within the field of supply chain management using MASs.

27 Multi-agent Systems

635

References Ahn HJ, Lee H, Park SJ (2003) A flexible agent system for change adaptation in supply chains. Expert Syst Appl 25(4):603–618 Barbati M, Bruno G, Genovese A (2012) Applications of agent-based models for optimization problems: a literature review. Expert Syst Appl 39(5):6020–6028 Browne J, Dubois D, Rathmill K, Sethi SP, Stecke KE (1984) Classification of flexible manufacturing systems. FMS Mag 2(2):114–117 van Brussel H, Wyns J, Valckenaers P, Bongaerts L, Peeters P (1998) Reference architecture for holonic manufacturing systems. Comput Ind 37(3):255–274 Caridi M, Cavalieri S (2004) Multi-agent systems in production planning and control: an overview. Prod Plan Control 15(2):106–118 Chaib-draa B, Müller J (2006) Multiagent based supply chain management. Springer, Berlin, Heidelberg Dilts DM, Boyd NP, Whorms HH (1991) The evolution of control architectures for automated manufacturing systems. J Manuf Syst 10(1):79–93 Douma AM (2008) Aligning the operations of barges and terminals through distributed planning. PhD thesis, University of Twente, Enschede, The Netherlands Douma AM, Schuur PC, Schutten JMJ (2011) Aligning barge and terminal operations using servicetime profiles. Flex Serv Manuf J 23(4):385–421 Ebben MJR (2001) Logistic control in automated transportation networks. PhD thesis, University of Twente, Enschede, The Netherlands Faratin P, Sierra C, Jennings NR (1998) Negotiation decision functions for autonomous agents. Robot Auton Syst 24(3–4):159–182 Franks NR (2009) Chapter 7 - Ants. In: Resh VH, Cardé RT (eds) Encyclopedia of insects, 2nd edn. Academic Press, San Diego, pp 24–27 Gerrits B, Mes MRK, Schuur PC (2016) Simulation of a multi-agent system for autonomous trailer docking. Beta Working Paper series 503 Hernández JE, Mula J, Poler R, Lyons AC (2014) Collaborative planning in multi-tier supply chains supported by a negotiation-based mechanism and multi-agent system. Group Decis Negot 23(2):235–269 Jennings NR, Faratin P, Lomuscio AR, Parsons S, Sierra C, Wooldridge M (2001) Automated negotiation prospects, methods and challenges. Int J Group Decis Negot 10(2):199–215 Lee HL, Padmanabhan V, Whang S (1997) Information distortion in a supply chain: the bullwhip effect. Manag Sci 43(4):546–558 Leitão P (2009) Agent-based distributed manufacturing control: a state-of-the-art survey. Eng Appl Artif Intell 22(7):979–991 Leitão P, Maˇrík V, Vrba P (2013) Past, present, and future of industrial agent applications. IEEE Trans Industr Inf 9(4):2360–2372 Leitão P, Vrba P (2011) Recent developments and future trends of industrial agents. Holonic MultiAgent Syst Manuf 6867:15–28 Liu SM, Gruver WA, Kotak DB (2002) Holonic coordination and control of an automated guided vehicle system. Integr Comput Aided Eng 9(3):235–250 Maturana F, Shen W, Norrie DH (1999) MetaMorph: an adaptive agent-based architecture for intelligent manufacturing. Int J Prod Res 37(10):2159–2173 Maturana FP, Norrie DH (1997) Distributed decision-making using the contract net within a mediator architecture. Decis Support Syst 20(1):53–64 Mcafee RP, McMillan J (1987) Auctions and bidding. J Econ Lit 25(2):699–738 Mes MRK, van der Heijden MC, van Harten A (2007) Comparison of agent-based scheduling to look-ahead heuristics for real-time transportation problems. Eur J Oper Res 181(1):59–75 Mes MRK, van der Heijden MC, van Hillegersberg J (2008) Design choices for agent-based control of AGVs in the dough making process. Decis Support Syst 44(4):983–999

636

M. Mes and B. Gerrits

Mes MRK, Iacob ME, van Hillegersberg J (2014) A distributed barge planning game. In: Meijer SA, Smeds R (eds) Frontiers in gaming simulation. Lecture notes in computer science, vol 8264. Springer, Cham, pp 214–221 Mitchell M (2009) Complexity: a guided tour. Oxford University Press, New York, NY, USA Monostori L, Váncza J, Kumara SRT (2006) Agent-based systems for manufacturing. CIRP Ann Manuf Technol 55(2):697–720 Müller JP, Fischer K (2014) Application impact of multi-agent systems and technologies: a survey. In: Agent-oriented software engineering: reflections on architectures, methodologies, languages, and frameworks. Springer, Berlin, Heidelberg, pp 27–53 Padgham L, Winikoff M (2005) Developing intelligent agent systems: a practical guide, wiley series in agent technology. Wiley, West Sussex, England Paolucci M, Sacile R (2016) Agent-based manufacturing and control systems: new agile manufacturing solutions for achieving peak performance. CRC Press, Boca Raton, FL, USA Parunak HVD (1987) Manufacturing experience with the contract net. In: Distributed artificial intelligence, pp 285–310 Parunak HVD (1998) Practical and industrial applications of agent-based systems. Industrial Technology Institute, pp 1–41 Parunak HVD, Baker AD, Clark SJ (2001) The AARIA agent architecture: from manufacturing requirements to agent-based system design. Integr Comput Aided Eng 8(1):45–58 Pˇechouˇcek M, Maˇrík V (2008) Industrial deployment of multi-agent technologies: Review and selected case studies. Auton Agent Multi-Agent Syst 17(3):397–431 Shen W et al (2006) Applications of agent-based systems in intelligent manufacturing: an updated review. Adv Eng Inform 20(4):415–431 Smith RG (1980) The contract net protocol: high level communication and control in a distributed problem solver. IEEE Trans Comput 29(12):1104–1113 Trentesaux D (2009) Distributed control of production systems. Eng Appl Artif Intell 22(7):971–978 Um W (2010) A study of multi-agent based supply chain modeling and management. iBusiness 2(4):333–341 Vis IFA (2006) Survey of research in the design and control of automated guided vehicle systems. Eur J Oper Res 170(3):677–709 Wooldridge M (2009) An introduction to multi-agent systems, 2nd edn. Wiley

Chapter 28

Artificial Intelligence Applications Matthias Klumpp

Abstract Automation, machine learning and predictive analytics all adhere to the same objective: improvement of operations, logistics, and supply chain processes by means of enhanced information technology applications and machines. This chapter outlines definitions and connections to operational and strategy questions in supply chain and logistics management for this new and evolving field of research as well as business application. Therefore, first basic developments and definitions of AI applications are described after which a case study is provided. Subsequently, applications of AI in intralogistics as well as core conceptual issues and problems are outlined (basic level). Applications in transportation and elaborate concepts and interactions with other management areas are discussed at a more advanced level. Finally, AI applications in supply chain design as well as an outlook for AI developments relating to logistics are described (state-of-the-art level), followed by suggestions for future topics and further reading material.

28.1 Developments and Definitions (Basic) Artificial Intelligence (AI) can be captured in a nutshell as “using machines to solve problems like humans”. The concept is quite old, more than 60 years as the term as well as the mission was coined at Dartmouth in 1956 by McCarthy and colleagues (Pan 2016). At that time, a new science field was envisioned for AI as “science and engineering of making intelligent machines” (Hamet and Tremblay 2017). But the core missions of making machines ‘understand, think, and learn in a similar way to human beings’ was in the first four decades blocked by many technical hardware and software hindrances—the time was just not ripe and therefore many expectations and projects were huge disappointments (Pan 2016). But in the past two decades, slow and increasing success is imminent in different fields, making it one of the most promising M. Klumpp (B) FOM University of Applied Sciences and University of Duisburg-Essen, Essen, Nordrhein-Westfalen, Germany e-mail: [email protected] © Springer International Publishing AG, part of Springer Nature 2019 H. Zijm et al. (eds.), Operations, Logistics and Supply Chain Management, Lecture Notes in Logistics, https://doi.org/10.1007/978-3-319-92447-2_28

637

638

M. Klumpp

Fig. 28.1 Development and definition of weak and strong AI

research fields today—as well as the single most important topic for venture capital and business investments of our time today: More than 200 AI companies were involved in VC (Venture Capitals) and M&A (Mergers and Acquisitions) deals since 2012, with more than 30 acquisitions taking place in the first quarter of 2017 (CB Insights 2017). For further analysis and connections towards logistics and supply chain management, it is informative to reconsider the literature distinction between “strong” and “weak” AI: Whereas strong AI would be able to solve a large number or even more tasks like human beings, weak AI is supposed to be able to solve one specific task similarly or better than humans, such as playing chess for example (see the left side in Fig. 28.1). As can be recognized from this example, weak AI is already a reality in our everyday lives when machines play chess, navigate cars, compose music or suggest shopping items as well as social contacts and places to go. Strong AI on the other hand is currently yet a vision for the future, recognizable roughly in major movies such as “Chappie”, “Automata” or “Ex Machina”—where AI robots are capable of complex bundles of activities such as understanding language, speaking, recognizing other persons and devising plans of their own actions—represented by the right-hand side of Fig. 28.1. This point1 of artificial intelligence reaching human intelligence levels is labeled “Artificial General Intelligence (AGI)”. The point in time where such AGI entities are reached is called “singularity” (Hanson 2008). After this point, we expect the entity featuring AGI to (self-)improve further, reaching a state of “Artificial Super Intelligence” (ASI)—which will be discussed further in Sect. 28.8 of this chapter. 1 It

is hard to define if it really will be a point in time, we may also expect a gradual transition with an ever-increasing number of AI applications e.g., in autonomous driving, autonomous behavior of robots as well as assisting devices e.g., in smartphones.

28 Artificial Intelligence Applications

639

But as the movie “Transcendence” shows vividly, we may not recognize an AGI entity at first when it encounters “first contact” with us humans as it may not come about in any humanoid form as the former film titles suggest—it is even unlikely, as unlikely as the expectation that life from outer space should have distinctive human forms, habits, and manners. This is also described by Armstrong in a “fighting anecdote” entitled “strength versus intelligence” of a dumb terminator (losing the battle by far) and a simple laptop device but with AGI powers—winning by luring, misleading, escaping via an internet connection established for it by the terminator in good faith and finally sending a cruise missile towards the terminator, read with Armstrong (2014, pp. 1–4). So, we are in for a surprise on this—let’s see and work on the weak notion of AI in the meantime. Furthermore, it has to be incorporated that “tasks” in the nutshell definition usually involve a complex bundle of information, analysis, decision, and activity processes. This is surely true for logistics processes such as order picking, route scheduling or truck driving. Therefore, usually AI innovations are happening in “jumps” and breakthroughs rather than as a continuous development—which again makes predictions in this field very hard and sketchy (Schmidhuber 2015). As indicated in Fig. 28.1, weak AI applications have increasingly been able to master single tasks or clearly restricted bundles of tasks such as playing chess or other accomplishments. This is indicated chronologically in Table 28.1 for some gaming examples—showing that, beside the fact that this list will grow longer and longer, the process also continues to accelerate. For the last line in this table taken from Bostrom (2014, pp. 12–13) we again recognize how hard future prognoses are in this field: Bostrom posits in 2014 that “go” as a very strategic and intuitive game may be mastered by AI on superhuman levels (meaning leaving human players no real chance as you may try with any free chess app on your smartphone today) “in about a decade”; this would indicate about 2024. But in reality, this level was already reached when the AI program AlphaGo beat the human go world champion Lee Sedol in March 2016—eight years earlier than predicted just 2 years before (Borowiec 2017). To lay the groundworks for later discussions as well as practical applications in the operations, logistics and supply chain domain, the following extended definition of AI is proposed (based on Bostrom 2014): • Artificial Intelligence encompasses a bundle of machine capabilities including the capacity to learn, the ability to deal with uncertainty and probabilistic information as well as the competence to extract useful concepts from sensory data and the ability to derive combinatorial representations for logical and intuitive reasoning; this leads to overall capabilities in several (and self-extending) interconnected fields of information analysis, decision making, communication and interaction – at the same (Artificial General Intelligence) or higher level as average humans (Artificial Super Intelligence).

In general, a series of expectations or expected advantages are connected to the development and implementation of AI applications in transportation, intralogistics, and supply chain steering. Such foreseen advantages of AI applications in logistics include—besides limitations and resistance areas discussed in Sect. 28.6:

640

M. Klumpp

Table 28.1 Advances in AI game applications (Bostrom 2014, pp. 12–13) Game Explanation Backgammon

1979: The backgammon program BKG by Hans Berliner defeats the world champion—the first computer program to defeat (in an exhibition match) a world champion in any game—though Berliner later attributes the win to luck with the dice rolls 1992: The backgammon program TD-Gammon by Gerry Tesauro reaches championship level ability, using temporal differences learning (a form of reinforcement learning) and repeated plays against itself to improve. In the years since, backgammon programs have far surpassed the best human players

Chess

1997: Deep Blue beats the world chess champion, Garry Kasparov. Kasparov claims to have seen glimpses of true intelligence and creativity in some of the computer’s moves. Since then, chess engines have continued to improve

Scrabble

As of 2002, Scrabble-playing software surpasses the best human players

Jeopardy!

2010: IBM’s Watson defeats the two all-time-greatest human Jeopardy! Champions, Ken Jennings and Brad Rutter. Jeopardy! Is a televised game show with trivia questions about history, literature, sports, geography, pop culture, science, and other topics. Questions are presented in the form of clues, and often involve wordplay

Go

As of 2012, the Zen series of go-playing programs has reached rank 6 dan (expert grade) in fast games (the level of a very strong amateur player), using Monte Carlo tree search and machine learning techniques. Go-playing programs have been improving at a rate of about 1 dan/year in recent years. If this rate of improvement continues, they might beat the human world champion in about a decadea

a Addendum:

AI application AlphaGo beat human world champion Lee Sedol in March 2016 (Borowiec 2017; Silver et al. 2016)

• Cost savings: AI applications are expected to achieve cost savings, especially in the area of personnel costs as robotics and automated intralogistics and transportation applications are expected to provide faster, higher quality, reliable, and durable—and therefore finally cheaper—solutions in the logistics domain. • Earnings increases: In many cases AI applications are expected also to increase revenue volumes—whether directly by increasing available product items at the point of sale (avoiding out of stock situations), or indirectly by allowing retailers to match customer preferences and expectations better and therefore increase customer satisfaction and re-buy rates. • Increased speed: In logistics and transportation contexts, an important dimension is speed, to be established by a more aligned cooperation of different actors within operations and transportation processes—e.g., between picking, production, packing and outbound transportation. This may significantly decrease lead times and time to market rates with the application of AI and automation. • Increased flexibility: Finally, especially with speedy AI applications, most logistics and SCM managers also expect increasing flexibility for intralogistics as well as

28 Artificial Intelligence Applications

641

Fig. 28.2 Example for advanced AI application expectations optimizing retail SKU and turnover (Glatzel et al. 2016, p. 3)

transportation setups; this is a very valuable asset in logistics environments and can be especially crucial in peak times.2 As an example for the expected cost savings as well as earnings increases by AI applications in supply chains, a detailed example from the retail sector is provided by Glatzel et al. (2016), and depicted in Fig. 28.2. The AI application addressed the classical “out of stock” problem and replenishment planning in retail and point of sale operations. By identifying the optimal solution regarding the specific replenishment lot size, the application is able to avoid lost earnings as well as too high cost levels for storage, transport and discarding unsellable items. The remainder of this chapter is organized as follows. At the basic level, Sect. 28.2 presents an instructive case study (Blue Yonder), Sect. 28.3 outlines applications of AI in intralogistics and Sect. 28.4 describes some basic principles and applications hurdles for AI in logistics. For the advanced level, Sect. 28.5 outlines the past and 2 Compare

for example www.magazino.eu—where individual picking robots can be transferred to other application areas as “swarm intelligence” which allows all robots to learn what one of them is experiencing; this increases flexibility in intralogistics tremendously.

642

M. Klumpp

current trends in assisted and autonomous driving, especially for trucks. Section 28.6 describes the current and future problems of human-computer interaction (HCI). On the state-of-the-art level, Sect. 28.7 describes application scenarios for AI in supply chain design, Sect. 28.8 outlines conceptual developments and Sect. 28.9 provides an outlook with trends and further reading options.

28.2 Case Study: Blue Yonder Retail Solution How OTTO Improved the Customer Experience by Using Predictive Applications OTTO, the German multichannel retailer based in Hamburg, successfully mastered the transition from classical mail-order retailer to online retailer by permanently adapting its business processes and by successfully reorienting its enterprise. Today, the online shop (www.otto.de) is the focus of the retailer’s business, accounting for 90% of its annual sales which amount to over 2.5 billion Euros (fiscal year 2015/16). One of the main reasons for this positive development is the company’s extensive range of products offered. Alongside fashion items and technical products, OTTO also sells furniture, sports articles, shoes, and toys. The online shop has a total of about 6,000 brands and more than 2.2 million article items. To withstand growing competitive pressure in e-commerce and to keep up with rapidly changing trends in the fashion industry, OTTO is using the latest in innovative technology. Multichannel retailer OTTO’s competitive environment is characterized by low margins, high competition, and increasingly faster changing market conditions and customer demands. Immense data volumes, a multiplicity of influential factors and a permanent need to act in real time under great time pressure are factors that typify decision-making processes. Positive customer experience and satisfied customers are at the heart of OTTO. To wow and retain customers, OTTO offers everything on one platform: a comprehensive selection of products at competitive prices with excellent service. The success factors here include the highest level of goods availability and short delivery periods. All these elements are intertwined, which presents a challenge. Only those who can deal with this will survive and be successful against the fierce competition. OTTO mastered these challenges together with Blue Yonder. Cloud-based predictive applications not only help the retailer improve customer experience, but also increase sales, reduce stock levels and decrease returns by basing strategic decisions on data. Blue Yonder develops its predictive applications with OTTO on an ongoing basis to be more practically oriented. The solutions adapt quickly and flexibly to circumstances in the increasingly changing market and make it possible for OTTO to achieve superior results with its business.

28 Artificial Intelligence Applications

643

(I) What is the optimal replenishment strategy? The fashion and lifestyle retail sector is a highly seasonal business. On the one hand, retailers must guarantee product availability for the entire season. On the other hand, there should be as little stock left in the warehouse as possible at the end of the season. To manage this balancing act and ensure business success at the same time, OTTO must be able to maintain precisely the optimal synergy between product availability and pricing for every single article in its extensive product portfolio. One of the greatest challenges in this is predicting the probable sales of an article at an early stage, because the profitable purchase of goods determines overall success. The most important task is to continuously ascertain the correct numbers. For many years, OTTO has been working closely with Blue Yonder to uphold its precise item-level sales forecasts. Blue Yonder’s algorithms were trained on historical data with a wide variety of input variables. With machine learning technology, the solution continuously evaluates its own forecasting quality and learns from past events. Today, Blue Yonder’s article item sales forecasts are a fixed part of the operative business processes at OTTO. On a daily basis for each article, an up-to-date forecast is made per color and size and based on hundreds of different input variables (i.e., brand, price, online placement, stock situation, weather). This means that OTTO provides Blue Yonder with millions of data records each week. Every year, more than five billion individual forecasts are created. Forecast quality improved by up to 40% per article compared to the conventional process and overstock was reduced by 20% at the end of the season. (II) What price is the customer willing to pay? Today, the requirements of intelligent price management are much greater than during the era of the catalog. The customer always expects a good price, and price transparency for brand name products is close to 100%. The basic question to ask for each item is: what price is the customer willing to pay? The optimal price for a product depends on many influencing factors that can vary on a daily basis: day of the week, season, time of day, weather, channel and device, competitors’ prices and much more. At any point in the product life cycle, there is an optimum price for a product. The challenge is to set it in relation to time. The right price at the right time increases customer satisfaction, leads to more sales and higher profit in the end. Experience shows that even the rate of returns decreases with optimal pricing. Today, Blue Yonder Price Optimization supports OTTO in successfully finding the “ideal” price. Blue Yonder Price Optimization for online retail examines and measures the connection between price changes and demand patterns. Based on a number of price-quantity pairs, Blue Yonder’s solution can pinpoint the price elasticity for every article item. Even for products that are difficult to sell, it can define the precise price elasticity by using cluster and collective algorithms and evaluating descriptive characteristics. Specific knowledge of the price elasticity makes it

644

M. Klumpp

possible to find the ideal price for a product according to the chosen price strategy. Based on the results, Blue Yonder automatically determines sales- or profit-increasing prices for the season. In a six-month pilot project in menswear, OTTO tested how it could automate pricing with the help of Blue Yonder’s solution, in order to increase sales and revenue. With impressive results: Blue Yonder was clearly in a position to significantly improve sales, revenue, and overall results—especially by reducing return rates by more than 50%. Based on this, the solution was also implemented in women’s apparel, and successive rollout is underway on the entire product range. Source: Blue Yonder (2017). (Video on https://www.blue-yonder.com/de/node/1812).

28.3 AI Intralogistics Applications (Basic) Automation in the intralogistics domain is discussed in Chap. 24. Therefore, the following short outline of practical AI application in this area will focus on new developments based on (a) independently acting robots with embedded AI in a more general sense, and (b) hybrid systems with AI and robotics developments helping human operators. (a) For the first application areas, Table 28.2 displays two cases of robots developed as current state-of-the-art. In both cases, applications in operations and logistics are conceivable but not yet fully in place. (b) The second area of AI application development is also closely linked to robots, but with an intelligent application—the hybrid robot-human-systems e.g., exoskeletons. A company engaged in this sector is SuitX (www.suitx.com)— with many potential applications in production and transportation. Applications in this area are expected to overtake stand-alone mobile robot applications in the near future (see Fig. 28.3). Compared to other application areas, industrial and logistics—e.g., picking and packing—use of exoskeletons faces a number of relevant challenges (based on Van de Venn 2016): • Exoskeletons are operated in a not clean and non-smooth environment; therefore, there exists the risk of staining and corruption of the device. • Typical movements are lowering, lifting and carrying over short distances (10 m

Low Frequency (LF) High Frequency (HF) Ultra High Frequency (UHF) Microwave.

This naming is general but the exact frequency ranges are governed by national regulations. Table 30.3 shows the permitted frequencies and frequency bands allowed in Europe along with the ranges that can be achieved. The quoted values in the Table 30.3 are theoretical. The actual ranges strongly depend on multiple factors such as the antenna/coil size, the transmission power, the presence of interference, and the surrounding environment. In practice, it is possible to achieve ranges of a few centimetres in the 125 kHz range and less than a metre in the 13.56 MHz frequency range. Furthermore, there are strict regulations governing the transmitting powers for the use of transponders as well. In addition to the frequency variations, different protocols are available for the communication using the RFID technology.

30.3.3 Bulk Reading and Collision Avoidance Within different stages in a supply chain process, a large number of items is stored or transported together on bigger units such as pallets or containers. However, at any checkpoint such as submission, delivery or customs clearance, identification of each single object is necessary. Therefore, an option to automatically read the identifiers in bulk is a major advantage. Since an RFID system does not require direct visual contact, it is potentially a proper candidate for bulk reading. However, special technical specifications and considerations are needed to enable implementation of bulk reading. Standard ISO 15693 lays the foundations for the bulk reading of transponders. An “inventory” command from a reader causes all transponders within the antenna’s range to respond with their respective identification numbers. The reader must detect and process the response telegrams from the individual transponders using a suitable collision-free method. Depending on the number of items on the transport object and their materials, the reader signal may be blocked. In such a case, RFID gates which have multiple operating readers from different angles to ensure reliable reading of

704

M. ten Hompel et al.

Fig. 30.10 An RFID gate (courtesy of TU Dortmund University, FLW)

all tags provide an appropriate solution. A schematic example of such an RFID gate is presented in Fig. 30.10. While all the transponders within the readers’ range will send their telegram directly after the inventory command, the transmission of the identification codes may lead to a collision of the telegrams. Therefore, a special algorithm is necessary to ensure a collision free read access of all tags. Collision is a familiar problem in both bus systems and wireless networks and there are multiple ways to address this issue. However, some of these methods are not applicable due to the limited technical capabilities of transponders. To solve the problem of collision in the field of RFID technology, we distinguish three (groups of) methods: ALOHA, Binary Tree (BT) and Query Tree (QT): ALOHA The ALOHA anti collision protocol itself has multiple variants, known as: • • • •

Pure ALOHA (PA) Slotted ALOHA (SA) Framed Slotted ALOHA (FSA) Dynamic Framed Slotted ALOHA (DFSA).

These variations are used in different standards and products such as ISO 18000-3, ISO 18000-6A, ISO 14443-3, EPCGlobal C1 and Philips I-Code. Generally speaking, they are based on acknowledgement protocols or time slotting procedures. However, a detailed discussion of the various implementations is beyond the scope of this chapter.

30 Automatic Identification Technology

705

Binary Tree (BT) Within the BT collision avoidance protocol, all tags have an internal counter initially set to zero. In case of a collision, each tag generates a random binary value. Tags with the binary value 1 increase the counter by one. In the next round, all tags with the counter value zero retry to transmit. In case of a collision, the procedure is repeated. In case of a successful transmission, all other tags reduce the counter value. This process is repeated until all tags have reached the counter value zero again. Query Tree (QT) The QT collision avoidance protocol uses a series of transmissions to narrow down the tag’s ID. At first, the reader transmits an EPC prefix to query the transponders. Among the available tags only those with a matching prefix will reply. Afterwards, the reader extends the prefix until it successfully reaches a single tag’s identifier. This process is repeated until no further reply is sent to the EPC prefix.

30.3.4 Fields of Application RFID is a general identification system with its own pros and cons which can be implemented in various application domains. Some of the most common applications are reviewed here. The use of RFID in industrial production may help to optimize the production process in parallel to the identification. It provides extra data on the timing which in turn is useful to support the production planning. Moreover, using RFID tags with internal memory will help to store the production data in a tag which travels with the item all the time and makes product tracking easier. In logistics, the use of transponders for labelling goods and packaging in the consumer goods sector offers considerable potential for optimising the management of the respective supply chains. The use of a bulk reading Auto-ID system such as RFID in the retail sector, may reduce the warehouse management and object identification efforts and hence logistics costs. In addition, object tracking is made much easier in a contactless manner. Price tagging and update can also be improved by contactless data transmission. Using RFID tags in airports helps to improve the tracking and tracing of the packages and hence improves baggage transmission. Still, due to the higher price of the RFID tags compared to barcode labels, implementation of RFID at airports is not so common. It is possible to use advanced RFID tags which are empowered with temperature sensors. They control the cooling chain of special products that are temperature sensitive such as butchery products, and medical and chemical goods. In case of a mismatch between the allowed temperature profile and the actual temperature, a failure is stored in the RFID tag and will be transmitted via the next reading process.

706

M. ten Hompel et al.

Personal authentication and access control is another application of RFID systems. In particular, a tag can be integrated in the personal identification of staff members and be read at the gates or entrances. RFID tags can be used for automatic toll applications as well. An antenna is then mounted on the wind shield of the cars, to be read by the readers mounted on the toll gates. This helps to improve traffic by reducing the time required for toll ticket purchase. 30.3.4.1

An Industrial Case Study

As mentioned, RFID technology is now integrated in a wide variety of applications. With an example of the automotive manufacturing industry some benefits of the technology are explained here. The current auto industry is largely dependent on its global components delivery and therefore any cost reduction in its supply chain is important. Note that the sector is characterized by a high individualisation of products which causes a large variety in component production. In modern cars, user electronic facilities include the entertainment system, navigation facilities, audio system, and connectivity. In spite of functionality and price differences in these systems, in most production lines these units look physically the same. Although using barcode will help to identify any single device, the integration of an RFID tag in each device provides much more efficiency in the production process. Since it is possible to read them in bulk, the time for scanning a pallet of these devices delivered to the assembly line is significantly reduced. Moreover, a barcode on the device would always require a central data bank to retrieve device specification, whereas all these specification can be stored on an RFID tag without the need for such a central data centre. Hence, using RFID tags leads to process decentralization and reduces the time and cost of communication and maintenance of a central data bank. In addition, it eliminates the need for data transfer from the producer to the assembly line. The possibility of rewriting an RFID tag multiple times enables to keep track of detailed information of all assembly steps in a decentralized manner. Moreover, this data is always accompanying the part and can be read without the need for accessing the central data bank. Another benefit of using RFID tags instead of barcodes in this sector is cost reduction by reducing the time for scanning the code. In a barcode based system, the code has to be affixed on a part such that it is visible for an automated barcode reader or an operator has to manually position the part properly. The exact location of the RFID tag is not that critical and the automated reading process is much more simple and reliable.

30 Automatic Identification Technology

707

Classical material flow control (organised hierarchically)

Internet of things (self-organising) Material flow (real time)

Servers Logistics object with RFID tag WMS clients Instantiation of agents Material flow computer

PLCs

RFID reader

RFID server

Material flow (real time)

RFID server

RFID server Servers

Fig. 30.11 From classical material flow control to the internet of things (courtesy of TU Dortmund University, FLW)

30.3.5 RFID in the Internet of Things (IoT) With the Internet of Things, it is possible to standardise the data on the tags and consequently, the material flow environments in which the smart logistics objects move. As opposed to a classical production planning and control system, control in an IoT-based production environment is far less hierarchical (Fig. 30.11). Indeed, within an IoT-based system, objects do not run through a pre-planned process chain. Instead, the logistics process is only created during runtime, based on standardised information exchange. Therefore, it is no longer possible to predict exactly how a control system based upon this concept will respond. The Internet of Things (IoT) is entirely based on RFID technology for controlling logistics systems. Its application addresses the challenges of in-house logistics in three different ways. Real world awareness In accordance with the concept of the Internet of Things, the individual logistics objects, i.e. the pallets, containers and parcels, are first equipped with RFID tags. The information needed to identify the objects is written to the tags. This prevents the often occurring mismatch between virtual inventory control and near real-time material flow control. People and machines can identify the “things” on-site by using a scanner to read all necessary information directly from the goods, enabled by the information storage capacity of the tag. As the information can be modified, it is also possible to permanently store a picking process on the container, for example. This does not contradict the concept of centralised data storage, but it allows to reconcile the stored data with reality on-site. The near real-time exchange of data is therefore performed in an entirely decentralised manner, while the logging, asset tracking and scheduling can

708

M. ten Hompel et al.

be executed centrally, as was already the case. This situation, which is referred to as real world awareness, permits the logical organisation of inventory management in in-house logistics. The database stores a snapshot of reality at a selected time point. Decentralisation In order to design a flexible and variable material flow system, it is necessary to decompose the system into modules. Without this modularisation, it is not possible to arrange individual components, elements and modules into new architectures. In the field of mechanical engineering, the modularisation of materials handling elements is also used to achieve standardisation and design of productive manufacturing and assembly processes. However, if consistent decentralisation is applied in order to achieve variability and flexibility, it is necessary for individual modules to have an ability to make decisions. Otherwise, it is not possible to modify the arrangement of materials handling modules such as sorting gates, junctions or accumulation conveyors, for example, without the need for to reprogram the material flow control. An on-site decision can only be made based on appropriate information. This information is carried by the tags in the IoT concept, so it is only necessary to parameterise the respective module. Self-organisation Consistent decentralisation allows for the creation of simple rules for controlling the materials flow. For this purpose, tags and thereby their linked logistics objects are provided with destination information and priorities. This allows individual materials flow modules to make independent decisions on-site. It allows things to find their way toward their destination regardless of the individual layout, which in turn can be adapted to suit the respective logistics requirements. This simplest form of the IoT creates a high throughput, but does not yet meet the requirements for punctuality, flexibility and adaptability. To do so, coordination of the individual in-house logistics processes is required which in turn demands further information be written to the tag. This information allows the on-site software to set parameters within the individual modules in real time. Modules with integrated intelligence cooperate with each other as a multi-agent system. An agent is instantiated in a uniform manner in the decentralised controller of the respective module. These agents communicate with their environment and with neighbouring agents. They facilitate the completion of a mission that is stored on the tags. For example, the agents can initiate priority rules, sequence the generation of picking orders and negotiate these with each other. For a more elaborate introduction on multi-agent systems, the reader is referred to Chap. 27. Such a modular multi-agent approach adds multiple advantages to the system compared to a classical system. Some of these benefits are: • Autonomy: agents operate autonomously without external manipulation • Social interaction: agents interact with the user and with other agents. Communication takes place on a semantic level, going beyond the execution of a set of instructions

30 Automatic Identification Technology

709

• Reactivity: agents perceive their environment and respond appropriately in a timely manner • Proactivity: agents not only respond to the environment but are also able to act in a goal-oriented manner and on their own initiative.

30.3.6 Selection of RFID As with any identification technology, the correct use of RFID technology calls for an analysis of the specific application domain circumstances. By analogy with the “six rights” rule in logistics, the six rights rule of radio frequency identification can be helpful. These rules for the selection of RFID technology are: 1. 2. 3. 4. 5. 6.

the right transponders (active, passive, UHF or HF, etc.) at the right location (case or item tagging) with the right data (EPC, EAN 128, data-on-tag or only identification, etc.) at the right point in the process (added value through quality and productivity) with the right middleware (integration into the rest of the system) at the right cost.

In general, one should always attempt to design process chains that are as complete as possible. The inclusion of an appropriate RFID technology may significantly help to reach that goal.

30.4 Future Automatic Identification Technologies (State-of-the-Art) This part briefly discusses problems, solutions, and new trends of future identification technologies. Next, a short overview of some ongoing research projects in this field is given.

30.4.1 On Metal Tag Systems An RFID tag cannot be easily attached on a metal surface or on any conductive surface. The metal surface hinders the RFID tag antenna in receiving signals from the reader as well as in transmitting signals to the reader (Ranasinghe et al. 2004). More precisely, if an RFID reader attempts to transmit energy to an RFID tag mounted on a metal surface, the emitted energy is reflected by the metal surfaces. The resulting interferences aggravate the communication between the RFID reader and RFID tag. There exist solutions that allow the RFID tag to work on metal surface; some of them are shown in Fig. 30.12. However, such solutions are too expensive to be of

710

M. ten Hompel et al.

Fig. 30.12 Some on metal RFID-Tags (courtesy of TU Dortmund University, FLW)

practical use for a broader scope of applications. Possible applications of today’s RFID on metals are, but are not restricted to: • Oil-pipelines • Gas pipelines • Vehicle tracking identification. The design of cost-efficient RFID tags embedded on a conductive surface such as metal (RFID on metal) is still an issue that needs to be solved by the RFID industry, in particular with a focus on improved communication and lower energy consumption. The main problem of designing RFID on metal tags has now been solved using metamaterials, such as Artificial Magnetic Conductor (AMC) (Feresidis et al. 2005). The drawbacks of such a solution are the risk of damage upon contact and the larger sizes of the RFID tag (Feresidis et al. 2005). Printable Metal-Mount RFID tags (see Fig. 30.13) are designed to be mounted directly on metal surfaces and incorporate an AMC. It is used to improve the communication quality between RFID tags and reader. The AMC substrate is used as a reflector for the tag antenna to mitigate the disturbance as well as to improve the communication quality at ultra-high radio frequency (UHF). However, the communication range is limited to a few meters only (Kim and Yeo 2008; Abdulhadi and Abhari 2012). Another solution to cope with metal surface is to use LF or HF RFID tags. This is the case if the read distance between the RFID reader and RFID tags plays only a minor role in the application. One further possibility to cope with metal or conductive surfaces depends on the design of a new antenna for RFID applications at UHF (Elsadek et al. 2013).

30 Automatic Identification Technology

711

Fig. 30.13 An example of on metal RFID-Tags (courtesy of Fujitsu)

A better antenna design may result in a better tag performance in terms of less energy consumption and a higher communication range (Wilas et al. 2009). Overall, on metal RFID solutions and metal-mount RFID tags are expected to play an important role in future solutions. The efficient design of metal-mount RFID tags is still a subject of further research and development.

30.4.2 Chipless RFID Systems To broaden the scope of RFID systems in IoT applications, new RFID tags have to be designed within a much lower price range. Progress in this direction has been made by eliminating the microelectronic chip in the transponder. This so-called chipless RFID tag is able to communicate with a reader at the UHF Band. The advantage of a chipless RFID tag is its lower price and its potential3 to replace the barcode label (Attaran and Rashidzadeh 2016). Therefore, it is expected that chipless RFID tags are going to be used in a variety of application domains. As an example, it is possible to design a chipless RFID tag for UHF applications by taking advance of Micro-Electro-Mechanical Systems (MEMS) technology, (cf. Attaran and Rashidzadeh 2016). The solution contains two configurable metal layers used to program either a unique backscattering pattern or an identification code in 3 Due

to lower production costs involved in manufacturing chipless RFID tag.

712

M. ten Hompel et al.

the RFID tag. According to Attaran and Rashidzadeh (2016), the prototype is able to communicate successfully with an RFID reader at the UHF band.

30.4.3 Authentication Protocols Due to the technological progress of RFID technology, not only the range of application domains becomes wider (IDTechEx 2015), but RFID tags are also expected to be used in areas with stringent requirements on data security (Thornton and Lanthem 2006). Therefore, data protection against eavesdropping and undetected alteration is required during the communication between the tag and the reader. Because of memory and processing power requirements of low-cost RFID tags, it is not feasible to use standard cryptographic algorithms (such as the Advanced Encryption Standard (AES)) to secure the communication between an RFID tag and the reader. Instead of giving up using authentication protocols (and consequently loosing possible application fields of RFID systems), appropriate cryptographic algorithms have to be designed. Hence, research has focused on the design of lightweight cryptography for low-cost RFID in recent years, such as: • • • •

Hash-lock (Chabanne et al. 2008) Link Aggregation Control Protocol (LCAP) (IEEE 2017) Lock-Free Cuckoo Hashing (Nguyen and Tsigas 2014) Lightweight Elliptic Curve Cryptography (ECC) (He et al. 2014)

Due to the widespread use of RFID systems in a large range of applications, the topic of security has become more and more important for the industry. Therefore, efficient lightweight cryptography solutions for low-cost RFID tags is also a subject of further research.

30.4.4 RFID Based Localization In addition to the identity of an object, its location plays an important role in several industry applications. For instance, indoor positioning techniques can be used by autonomous robots to plan or find optimal routes via which commodities are being transported. For outdoor applications, some good solutions with acceptable accuracy are already available to localize an object. The Global Positioning System (GPS), its Russian competitor GLONASS and the European solution named Galileo are three major existing outdoor platforms. These systems provide positioning information using satellite data and some complex algorithms integrated in the receiver device. However, the key drawback is the lack (or very weak reception) in indoor environments. Consequently, for indoor localization, other technologies have to be used.

30 Automatic Identification Technology

713

Although RFID is using radio based communication, neither active nor passive available systems (except some high-end expensive systems) are able to report the signal strength received from tags. This makes the location determination of tags roughly impossible. The only information provided by the current systems is the detection of presence in the field of their antenna which size depends on the frequency and the environmental conditions. However, the potential of localisation based upon RFID technology is available. There exist different approaches to estimate the location of an RFID tag such as distance estimation or Scene analysis. Algorithms based on distance estimation use properties of triangulation to estimate the position of an object, whereas scene analysis uses reference tags with known position to estimate the location of an object (Bouet and Santos 2008). More information on this topic can be found in Bouet and Santos (2008). Moreover, attempts are made to mount a grid of passive RFID tags in the floor, e.g. at the Technical University of Dortmund. When a reader travels on this floor, it will read the tags continuously and, based upon that, estimates the distance traveled. It can also be used as a time of flight system. Tags within reach of a reader’s antenna will reply to the reader request and those positioned at larger distance will require more time to transmit the data. Therefore, the reading frequency will be lower for them. In spite of the success of current methods, using RFID systems for the purpose of indoor localization remains an interesting and highly relevant field of research.

30.4.5 Near Field Authentication for RFID Technology One of the main issues for the Industry 4.0 or Smart Industry development is IT security. In Industry 4.0, all “intelligent” components of a factory such as machines and materials handling devices are connected to IT systems. They are able to communicate with each other using wired or wireless communication channels. If the communication among machines and other devices has to be protected, security is essential. IT security includes techniques to protect information and/or information systems against unauthorized access. This includes detection of unauthorized modification of information as well. Unprotected information exchange will increase the risk of undetected manipulation. Therefore, it is of utmost important that the information exchange (human-to-machine (H2M) and/or machine-to-machine (M2M) communication) in manufacturing technologies is protected. Using authentication is one of the fundamental methods for increasing security. Without using authentication, the information exchange between two devices cannot be protected against forging. Consequently, in the presence of an adversary, no guarantees can be given that an identifier belongs to a certain device or not, regardless the attachment of a unique and unambiguous ID to the device. In order to secure the

714

M. ten Hompel et al.

information exchange between two or more devices as well as to prevent unauthorized access to a device, authentication is a necessity. Following Cankaya (2011), the term authentication is defined as follows: Authentication is the process of verifying an entity’s identity, given its credentials. The entity could be in the form of a person, a computer, a device, a group of network computers, etc.

Consequently, authentication comprises the following tasks: Identification: in which an entity claims an identity (e.g. by transmitting a password) to get access or submit information to a source, Verification: in which the identity of the entity is verified too. For example, grant access to a source. This means that authentication in near or far field communication is used to ensure that electronic devices, such as an RFID tag which is attached to a product, are distinguished from others. When the ID of a device is not protected against forging, an adversary is able to replace a product x by another (“faked”) product y. This may result in system failure, quality issues, and financial losses. It must be clear that an ordinary and unprotected RFID tag with a unique ID is not sufficient to ensure the authentication. This holds true even when the communication between two or more devices (e.g. RFID tag and reader) is established within short ranges (e.g. 50

CO2 emissions (French electricity production mix)

−60

Traveled distance to reach destination Delivery lead-time (transportation only)

−15 −12

Inventory level (same service level)

−40

31.4 Start-up: A Case Study (Basic) The Physical Internet may look somehow disruptive compared to existing logistics operations. How to start? From where? Several start-ups have already proposed solutions inspired by the concept. Several domains are quite active: business language, software, communication tools, handling equipment and marketplaces. The focus is put here on a new type of logistics service, routing with a Collaborative Routing Center or CRC in short. The CRC proposes to route pallets of FMCG products from many suppliers to many retailers. A direct shipment takes place from the warehouse of each supplier to each distribution center. When it makes sense, several shipments for several retailers are dynamically bundled from a single (or several) manufacturers to the CRC. The CRC, for a fixed per unit transshipment fee, sorts units received and fills trucks towards the retailers. The CRC could operate near the suppliers or near the distribution points or both. If the foreseen fill rate of a transportation means is not sufficient for a shipper for a direct shipment, customers are added until full capacity is reached. If a shipper for a given destination reaches a full truckload, the CRC is skipped. The CRC is therefore a pay per use pooling solution utilized by shippers when it is in their best interest without any change in their supply chain, with one exception: the order quantity per customer (Fig. 31.5). The CRC started operations in the southeast of France near Marseilles at the end of 2015 after several months of experimentation. A different logistics services provider can operate each CRC as long as it is compliant with rules defined in an operational charter.

31 The Physical Internet

725

Fig. 31.5 A single CRC between origins and destinations with a consolidation factor of 3 (each lane contains 3× the load of the initial network with 3 × 3 lanes)

In 2016, six new CRCs were opened providing national coverage in France. One major key performance indicator of the CRC is the truckload. In 2016 with limited flows and customers, the fill rate was on average 87%—a major improvement compared to previous LTL or FTL operations. The next steps are: extension of the geographic coverage and development of new functionalities such as reverse logistics or storage and multimodal operations. One of the investors in CRCs operates a multimodal network based on swap containers as shown in Fig. 31.6. This new service in conjunction with biogas trucks, operated in the same network, has the potential to drastically reduce GHG emissions.

31.5 Key Components of the Physical Internet (Advanced) To interconnect two logistics services many aspects are involved. The emphasis is put here on three main components: physical with containers, information sharing and marketplace mechanisms.

726

E. Ballot

Fig. 31.6 Moving a swap container from the train to a truck

31.5.1 Physical Aspect: Containers In an “open” network where many services are reachable, it is important to protect the goods, business confidentiality and improve handling productivity. The maritime container provides a very inspiring example for inland logistics. With its standardized size, its unique identification number and the twist lock handling system, the maritime container revolutionized maritime shipping. Figure 31.7 shows the cost reduction especially in port transit and handling coming for its standard design. While about 2 sizes are enough to cover the vast majority of maritime needs, the situation is a bit more complex inland. With different vehicle sizes and the need to ship quantities, two types of containers of modular dimensions were identified: transportation containers and handling containers. Transportation containers, like maritime containers, ensure the interface with the transportation mean and protection against external elements. The handling containers look more like boxes with specific features: modular dimensions, sealable, attachable between themselves to form blocks and with a unique identification code (Montreuil et al. 2015). Several prototypes were designed in the European project Modulushca. As a consequence, the handling container set offers smaller shipment, does not require pallets for handling (by clamping) and enables footprint “free” operations (the block composition can be adjusted to reach the desired footprint). Figure 31.8 illustrates the principle. The definition of handling container set will be done sector by sector with industrials to fulfill their needs according to the design principles. The subject is now in the hands of the industry and their representatives.

31 The Physical Internet

727

Fig. 31.7 Comparison of shipment cost for the equivalent of a truck in 1960, in 2010 without innovation and the actual cost with the maritime container. Data from the author and Levinson (2006)

Fig. 31.8 Composition of handling containers to form a block

31.5.2 Information Sharing In an interconnected network, a shipment encapsulated in a handling container or a set of handling containers will be passed from one logistics service provider to another and potentially several times to optimize routing. This openness of the network and sealed handling containers reinforce the need for full digital visibility and control from authorized stakeholders at the level of containers to perform logistics decisions but limited to them.

728

E. Ballot

In the field of logistics, the goal is to be able to locate and communicate with the container to transmit the best decision about the next steps to be performed. The Internet of Things developments are of particular interest for the Physical Internet for two main aspects. Smart objects with sensors and communication capabilities and information definition readily facilitate exchanges between business partners. If one looks at technologies, we see very interesting developments in both sensors (accelerometers, temperature, localization) and long-range and near-field communication devices and networks at prices that are more and more affordable for logistics operations. In the near future supply chains will be full of smart objects such as containers, trucks, forklifts, conveyors etc. Figure 31.9 illustrates a single communication scheme from handling containers with the EPC global standard (GS1 2013). Several languages are necessary to efficiently exchange all the information necessary to coordinate activities among the actors to perform the various types of logistics services.

31.5.3 Routers Container routing in the Physical Internet is based on a physical level (a physical hub) and a routing level (a market place), supported by information sharing. At the physical level, the hub performs all physical operations (unloading of transportation means, inspection, decomposition, storage, sorting, composition, or shipping. It could be done by actual means (such as cross-dock platforms) or by new types of facilities aligned with Industry 4.0 principles (automated, resilient, flexible and scalable). A typical example is given by the grid flow concept from Gue et al. (2014), now operational with the FlexConveyor solution. Figure 31.10 shows the concept, a grid composed of bidirectional conveyor units with the same software in each unit. The grid is one possible expendable configuration among a huge number of potential configurations. A box can enter the grid from anywhere and could be extracted from anywhere according to the needs and can be stored anywhere on the grid. The routing level provides the best allocation of services to fulfill all commitments (transit time, delivery date or allowed expenditure). This intermediation service is very similar to a market place but with extended capabilities compared to traditional freight marketplaces. It combines several shipments to the best capacity offers thanks to the interconnection of services and the physical hub. Figure 31.11 gives an example where a hub is represented by a triangle and a shipment is represented by a request r. In this particular example, r 3 and r 4 are already bundled and allocated to a carrier t 1 . Carriers are noted from 1 to 4 according to their cost, t 1 is the cheapest and transshipment between carriers is authorized. They have all the same capacity of 10 load units. The best allocation for this problem is: t 1 transports r 4 , r 6 and r 7 , t 2 transports r 3 and r 5 , t 3 transports r 1 and r 2 and t 4 the most expensive transports nothing. Here request 3 is moved from carrier 1 to 2 while carrier 1 receives requested 6 and 7, see all

31 The Physical Internet

729

Fig. 31.9 Example of EPC global architecture with interface to legacy systems and information publication via several EPCIS that enables new services

details in Xu (2013). The optimal allocation is reached by an auction mechanism with independent players (carriers) and not a centralized planning. Of course, this example is very simple (few requests, few carriers, one hub, one period, no transshipment fee) but it illustrates the principle and let envision extensions. The idea is to generalize this kind of approach to ensure the optimization of both transportation means and routing of shipments. Chapter 21 presents more elaborate consolidation algorithms.

730

E. Ballot

Fig. 31.10 A configuration of the GridFlow, entries in blue, exits in black, storages in yellow, conveyor units in grey

r3

Request at hub 0 r1 r2 r3 r4 r5 r6 r7

Dest. 1 3 2 5 4 5 5

Load units 5 5 5 4 5 3 3

O

r4

10 r1

5

2

r7 10

1 r5 r2

5

10

5

3

t1

r6

4

Fig. 31.11 An optimal allocation of carriers for a set of requests in hub 0

31.6 Operations with the Physical Internet (State-of-the Art) The Physical Internet is the interconnection of all logistics services. As such the goal is not only to improve operational efficiency but also to reconfigure networks to support new organizations and services. Two examples are described below. The first example, illustrated in Fig. 31.12, shows how the Physical Internet could be used to ship directly and faster to the consignees as in parcel networks. But it could also allow operations corresponding to a classical distribution network such as picking (at container level), storage and flexible allocations.

31 The Physical Internet

731

130

Km / h

100 80 60 40 20 0

Without points of sales

0

2

4

6

8

10 12 14 16 18 20 22 24 26 Days

Fig. 31.12 Comparison of the classical SC organization and the Physical Internet organization

The PI also permits to consolidate shipments from many shippers before delivery according to services criteria such as: minimize GHG emissions, minimize costs, or any combination of criteria. The second example, Fig. 31.13, explores stocking strategies. In contemporary supply chains, the allocation of assets to a dedicated service or customer minimizes the number of stocking points. This is why most warehouses are located, based on a centroid approach, close to each other but far from customers. In a more shared and distributed approach, warehouses could be distributed more efficiently over a territory and could offer several stocking points to all manufacturers and retailers. The ability to store in several places could be exploited by inventory management algorithms to support multi-sourcing and inventory repositioning but more importantly a distribution (or a supply) network customized to each product characteristic. Such inventory management decisions are completely out of reach in current single dedicated networks. A completely distributed and shared logistics holds a potential not only for innovation but also to deploy services at a much faster pace compared to nowadays, where lack of standards act as barriers to innovations as they have to adapt to specificity of all services.

732

E. Ballot

Source of o supply: blue dot Warehoouses or hubs: green dot Point off demand: red dot d Numberr of orders: on yellow y dots

Notationn

Classiical distribution network: a ccommon sourcee of supply and d two independdent distributiion networks

Expensiive products, with w low demannd volume an nd high demand uncertainty: direct shipmennts (inventory in n one place)

Cheap p products, hig gh demand voluume and high demand unceertainty: shipm ments to hubs with multi-sourrcing and repossitioning of pro oducts (inventorry in several plaaces)

Fig. 31.13 Comparison of inventory management strategies (Yang et al. 2016)

31.7 Roles of Key Actors in the Design of the Physical Internet (State-of-the Art) The Physical Internet proposes a comprehensive approach to redesign logistics networks and supply chain operations. It will not be done overnight, neither switched

31 The Physical Internet

733

Fig. 31.14 The ALICE roadmap towards the Physical Internet with European milestones

on in one phase. It requires a significant design effort from researchers, industrialist, governments and institutions for a gradual implementation. The first level is awareness and education, it is achieved throughout several channels: journals, books, videos, and conferences. The second level in the enrolment is the design of the Physical Internet. Here several associations could play a significant role. The leader is ALICE, the Alliance for Logistics Innovation through Collaboration in Europe the European Technology Platform for Logistics http://www.etp-logistics.eu. Created in June 2013, ALICE attracted more than 100 major players in the design process of a roadmap towards the Physical Internet. Figure 31.14 gives the 5 main domains of research and the horizons based on the European research agenda. Other institutions such as GS1 (information) or the Consumer Goods Forum and Material Handling Industry (container design and handling) (Material Handling Industry 2014) are also involved in the process as well as public institutes in Asia (Hong Kong, South Korea and China). However, start-ups from the most dynamic area, with several solutions already on the market for stocking (Flexe.com, Stockbooking.com etc.) enable warehouses as a service, for transportation with many new marketplaces for shippers (Uship and others), in software edition with MixMoveMatch and others. The solutions might come from their side sooner, with an amplified impact if they are “interconnection ready”. The disruption in distribution channels and retailing is already well underway with new giants imposing new practices. They force all actors to adapt to new market rules. With this regard, the Physical Internet appears as a promising and fair alternative to the usual “winner takes all”.

734

E. Ballot

References Ballot E, Fontane F (2007) Transport performance and efficiency: overall vehicle effectiveness. In: CGS Working paper. Mines ParisTech, Paris Ballot E, Montreuil B, Meller RD (2014) The Physical Internet: the network of logistics networks. La documentation Française Essen H, Schroten A, Otten M, Sutter D, Schreyer C, Zandonella R, Maibach M, Doll C (2011) External costs of transport in Europe. CE Delft, INFRAS, Fraunhofer, Delft. ISI: 161 European Commission (2011) A roadmap for moving to a competitive low carbon economy in 2050. Communication from the Commission to the European Parliament, The Council, The European Economic and Social Committee and the Committee of the Regions. Office of the European Union, Brussels, 16 pp European Environment Agency (2017) Load factor for freight transport. http://www.eea.europa.e u/data-and-maps/indicators/load-factors-for-freight-transport/load-factors-for-freight-transport1. Accessed Feb 20171 Gaubert E, Guerrero D (2013) Les enquêtes Chargeur: l’observation de la demande de transport de marchandises au service des politiques publiques. IFSTTAR GS1 (2013) EPC Global Standards. http://www.gs1.org/gsmp/kc/epcglobal. Accessed Oct 2013 Gue KR, Furmans K, Seibold Z, Uludag O (2014) GridStore: a puzzle-based storage system with decentralized control. IEEE Trans Autom Sci Eng 11:2, 429–438 International Energy Agency (2009) Transport, energy and CO2, moving toward sustainability. I. Statistics. IEA, Paris International Transport Forum (2016) Capacity to grow. Transport infrastructure needs for future trade growth. OECD, Paris, p 49 Joumard R (1999) Methods of estimation of atmospheric emissions from transport: European scientist network and scientific state-of-the art COST 319 final report. INRETS, Bron, p 158 Levinson M (2006) The box. Princeton University Press, Princeton Material Handling Industry (2014) Material handling & logistics U.S. roadmap. MHI 67 McKinnon A, Ge Y, Leuchars D (2003) Analysis of transport efficiency in the UK food supply chain. In: Centre LR (eds). Edinburgh, p 38 Mervis J (2014) The information highway gets physical. Science 344(6188):1104–1107 Montreuil B, Ballot E, Tremblay W (2015) Modular design of Physical Internet transport, handling and packaging containers. International Material Handling Research, MHI, Charlotte, NC, p 2015 Sarraj R et al (2014) Interconnected logistic networks and protocols: simulation-based efficiency assessment. Int J Prod Res 52(11):3185–3208 Xu X (2013) Collaboration mechanism in horizontal logistics collaboration. PhD dissertation 2013, Mines ParisTech, Paris, p 209 Yang Y, Pan S, Ballot E (2016) Innovative vendor-managed inventory strategy exploiting interconnected logistics services in the Physical Internet. Int J Prod Res 1–18. https://doi.org/10.1080/0 0207543.2016.1275871

Smile Life

When life gives you a hundred reasons to cry, show life that you have a thousand reasons to smile

Get in touch

© Copyright 2015 - 2024 AZPDF.TIPS - All rights reserved.